SlideShare a Scribd company logo
1 of 49
Download to read offline
State	
  of	
  the	
  Web	
  -­‐	
  Q4	
  2009
A	
  View	
  of	
  the	
  Web	
  From	
  an	
  End	
  User’s	
  Perspec:ve

Zscaler	
  Labs




Abstract
Attackers	
  are	
  no	
  longer	
  targeting	
  web	
  and	
  email	
  servers.	
  Today,	
  they	
  are	
  attacking	
  
enterprises	
  from	
  the	
  inside	
  out,	
  by	
  <irst	
  compromising	
  end	
  user	
  systems	
  and	
  then	
  
leveraging	
  them	
  to	
  gain	
  access	
  to	
  con<idential	
  data.	
  As	
  such	
  it	
  is	
  imperative	
  that	
  
organizations	
  have	
  an	
  understanding	
  of	
  what	
  is	
  happening	
  on	
  the	
  web.	
  As	
  a	
  
Security-­‐as-­‐a-­‐Service	
  vendor,	
  Zscaler	
  has	
  a	
  unique	
  perspective	
  on	
  web	
  traf<ic.	
  With	
  
millions	
  of	
  end	
  users	
  traversing	
  the	
  web	
  through	
  Zscaler’s	
  global	
  network	
  of	
  web	
  
gateways,	
  we	
  are	
  able	
  to	
  better	
  understand	
  both	
  how	
  users	
  are	
  interacting	
  with	
  
web	
  based	
  resources	
  and	
  how	
  attackers	
  may	
  be	
  targeting	
  end	
  users.	
  In	
  this,	
  our	
  
<irst	
  quarterly	
  ‘State	
  of	
  the	
  Web’	
  report,	
  we	
  provide	
  a	
  window	
  into	
  the	
  web	
  from	
  an	
  
end	
  user’s	
  perspective.
Table	
  of	
  Contents
Overview	
  .....................................................................................................................................3
Web	
  Traf/ic	
  Statistics	
  ..............................................................................................................3
    Web	
  Server	
  Statistics	
  .......................................................................................................................3
      TLDs	
  by	
  Unique	
  Domain	
  Visited	
  ...............................................................................................................4
      TLDs	
  by	
  Total	
  Transactions	
                 .........................................................................................................................6
      Transaction	
  to	
  Domain	
  Ratio	
  .....................................................................................................................6
      Top	
  Domains	
  Visited	
  ......................................................................................................................................8
      CIDR	
  Block	
  Distribution	
  ...............................................................................................................................9
      ASN	
  Distribution	
  ...........................................................................................................................................11
      Geography	
         ........................................................................................................................................................12
      File	
  Types	
  .........................................................................................................................................................15
      Request	
  Method	
  ............................................................................................................................................15
      Response	
  Code	
  ...............................................................................................................................................16
    Web	
  Browser	
  Statistics	
               ..................................................................................................................17
      Browser	
  Version	
            ............................................................................................................................................17
    User	
  Statistics	
  ..................................................................................................................................19
      URL	
  Categorization	
  ......................................................................................................................................19
      Search	
  Engines	
  ...............................................................................................................................................19
      Social	
  networking	
  .........................................................................................................................................20
      File	
  Sharing	
  ......................................................................................................................................................20
      Government	
  ....................................................................................................................................................21
      Retail	
  ..................................................................................................................................................................21
Security	
  Statistics	
  ..................................................................................................................22
    Threats	
  ...............................................................................................................................................22
      Malware	
  By	
  IP	
  Address	
  ..............................................................................................................................22
      Malware	
  by	
  Country	
  ....................................................................................................................................22
      Phishing	
  ............................................................................................................................................................23
      Malicious	
  Domains	
  .......................................................................................................................................24
      Anonymizers	
  ...................................................................................................................................................25
      Botnets	
  ..............................................................................................................................................................25
    Traf/ic	
  ..................................................................................................................................................26
      Bogon	
  IP	
  space	
  ...............................................................................................................................................26
Conclusion	
  ................................................................................................................................28
Appendix	
  ..................................................................................................................................29
    TLD	
  Breakdown	
  ..............................................................................................................................29
      Monthly	
  Summary	
  –	
  Top	
  TLDs	
  Visited	
  ................................................................................................29
      Monthly	
  Summary	
  -­‐	
  Unique	
  Domains	
  Per	
  TLD	
  ................................................................................30
    Categorization	
  Breakdown	
  ..........................................................................................................31
      .COM	
  Breakdown	
  by	
  Category	
  ................................................................................................................31
      .NET	
  Breakdown	
  by	
  Category	
  .................................................................................................................34
      .ORG	
  Breakdown	
  by	
  Category	
  .................................................................................................................37
      .INFO	
  Breakdown	
  by	
  Category	
  ................................................................................................................40
    Top	
  Search	
  Queries	
  .........................................................................................................................49
Overview
Our	
  goal	
  in	
  producing	
  this	
  report	
  is	
  to	
  better	
  understand	
  traf<ic	
  on	
  the	
  web	
  today.	
  
Security	
  and	
  IT	
  teams	
  across	
  organizations	
  are	
  tasked	
  with	
  managing	
  the	
  traf<ic	
  of	
  
end	
  users	
  on	
  their	
  networks.	
  That	
  can	
  involve	
  restricting	
  access	
  for	
  various	
  business	
  
purposes	
  and	
  protecting	
  end	
  users	
  from	
  external	
  threats.	
  This	
  is	
  a	
  tall	
  order	
  given	
  
the	
  increasing	
  ease	
  of	
  access	
  to	
  web	
  based	
  services	
  that	
  often	
  permit	
  users	
  to	
  bypass	
  
traditional	
  controls	
  -­‐	
  whether	
  accessing	
  corporate	
  resources	
  from	
  personal	
  devices	
  
such	
  as	
  smart	
  phones	
  or	
  setting	
  up	
  applications	
  by	
  leveraging	
  cloud	
  based	
  
resources.

Zscaler	
  is	
  in	
  a	
  unique	
  position	
  to	
  observe	
  trends	
  in	
  web	
  traf<ic.	
  As	
  a	
  Security-­‐as-­‐a-­‐
Services	
  vendor,	
  Zscaler’s	
  network	
  of	
  web	
  gateways	
  continually	
  inspects	
  traf<ic	
  for	
  
millions	
  of	
  end	
  users	
  around	
  the	
  globe.	
  

There	
  are	
  a	
  number	
  of	
  great	
  reports	
  available	
  today	
  from	
  a	
  variety	
  of	
  organizations	
  
to	
  help	
  us	
  better	
  understand	
  web	
  traf<ic.	
  However,	
  the	
  majority	
  of	
  such	
  reports	
  tend	
  
to	
  focus	
  on	
  the	
  server	
  side	
  of	
  the	
  equation.	
  They	
  tend	
  to	
  look	
  at	
  the	
  technology	
  that	
  
has	
  been	
  deployed	
  to	
  deliver	
  web	
  content	
  and	
  associated	
  security	
  issues	
  in	
  web	
  
applications.	
  We	
  feel	
  that	
  there	
  is	
  a	
  need	
  to	
  better	
  understand	
  the	
  client	
  side	
  of	
  the	
  
equation	
  -­‐	
  what	
  are	
  end	
  users	
  doing	
  on	
  the	
  web	
  and	
  how	
  are	
  attackers	
  targeting	
  
them?	
  The	
  latter	
  part	
  of	
  this	
  question	
  is	
  especially	
  important	
  as	
  attackers	
  have	
  
clearly	
  shifted	
  away	
  from	
  attacking	
  web	
  and	
  email	
  servers	
  to	
  targeting	
  end	
  users.	
  
They	
  understand	
  that	
  end	
  user	
  systems	
  tend	
  to	
  represent	
  the	
  weakest	
  link	
  in	
  the	
  
security	
  chain	
  and	
  they	
  are	
  exploiting	
  that	
  weakness	
  with	
  increasing	
  ef<iciency.	
  We	
  
can	
  better	
  defend	
  against	
  such	
  attacks	
  by	
  better	
  understanding	
  exactly	
  what	
  is	
  
occurring	
  on	
  the	
  web	
  and	
  it	
  is	
  our	
  hope	
  that	
  this	
  report	
  will	
  help	
  to	
  shed	
  some	
  light	
  
on	
  that	
  very	
  topic.


Web	
  Traffic	
  Sta0s0cs
Web	
  Server	
  Sta0s0cs
Zscaler	
  customers	
  visited	
  several	
  million	
  web	
  servers	
  during	
  the	
  4th	
  quarter	
  of	
  
2009.	
  One	
  interesting	
  technique	
  for	
  visualizing	
  the	
  IP	
  addresses	
  of	
  the	
  web	
  servers	
  
visited	
  is	
  through	
  a	
  heatmap.	
  The	
  below	
  graphic	
  was	
  generated	
  from	
  the	
  
Measurement	
  Factory	
  software1	
  and	
  “uses	
  a	
  12th-­‐order	
  Hilbert	
  curve2	
  to	
  represent	
  
the	
  entire	
  IPv4	
  address	
  space”.	
  In	
  the	
  graphic	
  below,	
  IP	
  addresses	
  visited	
  are	
  
represented	
  by	
  white	
  pixels,	
  while	
  addresses	
  not	
  visited	
  are	
  displayed	
  as	
  black	
  
pixels.	
  Non-­‐routable	
  or	
  reserved	
  space	
  is	
  identi<ied	
  in	
  gray	
  and	
  where	
  appropriate,	
  
we	
  have	
  indicated	
  what	
  that	
  space	
  is	
  used	
  for.	
  It’s	
  a	
  fascinating	
  view	
  which	
  exposes	
  
just	
  how	
  vast	
  the	
  Internet	
  truly	
  is.	
  Even	
  when	
  analyzing	
  traf<ic	
  from	
  millions	
  of	
  users	
  


1   http://maps.measurement-factory.com/software/ipv4-heatmap.1.html
2A Hilbert Curve is a space filling curve that visits every point in a grid (in our case a 2^12 x 2^12
grid).
over	
  the	
  course	
  of	
  three	
  months,	
  it	
  can	
  be	
  seen	
  that	
  much	
  of	
  the	
  Internet	
  remains	
  
untouched.




Hilbert	
  Curve	
  -­‐	
  All	
  Q4	
  2009	
  traffic	
  by	
  IP	
  address

TLDs	
  by	
  Unique	
  Domain	
  Visited
                                                        .com,	
  .org,	
  and	
  .net	
  top-­‐level	
  domains	
  (TLDs)	
  
                                                        consistently	
  made	
  up	
  the	
  bulk	
  of	
  the	
  unique	
  
             Other
             10%
                                                        domains	
  visited	
  each	
  month.	
  .com	
  traf<ic	
  made	
  up	
  
     org                                                80.11%	
  of	
  the	
  unique	
  domains	
  visited	
  during	
  the	
  
     4%
   net                                                  quarter.	
  .net	
  had	
  over	
  4.96%	
  and	
  .org	
  accounted	
  for	
  
   5%
                                                        4.45%.	
  The	
  chart	
  below	
  shows	
  the	
  next	
  10	
  largest	
  
                                                        TLDs	
  that	
  make	
  up	
  about	
  80%	
  of	
  the	
  remaining	
  
                                                        11%	
  of	
  the	
  unique	
  domains	
  visited	
  in	
  Q4	
  of	
  2009.



                                  com
                                  80%
Top 10 TLDs By Unique Domain Per Month (Excluding .com/.net/.org)




                     0%                    0.50%                      1.00%                         1.50%
                                                                                                                                   2.00%
      1.25%
ru 1.1%
      1.50%
       1.04%
uk 1.5%
       1.22%
       1.03%
 au 1.3%
        1.30%
        1.23%
edu 1.3%
         1.11%
         0.74%
  de 1.0%
         1.07%
          0.59%
 info 0.6%
          0.63%
          0.63%
   us 0.6%
           0.57%
           0.56%
     fr 0.6%
           0.49%
            0.63%
     in 0.5%
            0.47%
            0.39%
     ca 0.5%
             0.44%




                           October                       November                          December

There	
  were	
  a	
  number	
  of	
  similarities	
  with	
  a	
  few	
  <luctuations	
  within	
  the	
  top	
  10	
  TLDs	
  
with	
  unique	
  domains	
  visited	
  from	
  month-­‐to-­‐month:
      •    .ru	
  was	
  the	
  4th	
  most	
  popular	
  TLD	
  by	
  unique	
  domain	
  visited	
  in	
  October	
  and	
  
           December,	
  however	
  it	
  dropped	
  to	
  the	
  7th	
  spot	
  in	
  November.
      •    .au,	
  .uk,	
  and	
  .edu	
  make	
  up	
  the	
  5-­‐7	
  spots,	
  with	
  the	
  exception	
  of	
  November	
  
           when	
  .uk	
  beat	
  out	
  .ru	
  for	
  the	
  4th	
  spot.

The	
  chart	
  below	
  shows	
  the	
  breakdown	
  of	
  TLDs	
  based	
  on	
  total	
  number	
  of	
  
transactions	
  as	
  opposed	
  to	
  unique	
  domains.	
  This	
  view	
  would	
  favor	
  those	
  TLDs	
  
hosting	
  popular	
  sites	
  which	
  receive	
  higher	
  volumes	
  of	
  overall	
  traf<ic.
TLDs	
  by	
  Total	
  Transac0ons
                Top 10 TLDs By Transaction Per Month (Excluding .com/.net)




                         0%                 0.50%                  1.00%                    1.50%                      2.00%
      1.57%
org    1.54%
       1.44%
       0.66%
  au 0.97%
        0.78%
        0.62%
   in 0.38%
         0.41%
         0.36%
    tv 0.35%
          0.55%
          0.29%
    uk 0.44%
          0.41%
           0.16%
    de 0.23%
           0.26%
            0.22%
   gov 0.19%
            0.25%
            0.24%
      fr 0.21%
             0.18%
             0.21%
     edu 0.18%
             0.20%
              0.23%
      pe 0.22%
              0.16%




                        October                          November                             December

Transac0on	
  to	
  Domain	
  Ra0o
The	
  data	
  from	
  the	
  top	
  TLDs	
  by	
  unique	
  domain	
  and	
  top	
  TLDs	
  by	
  transaction	
  can	
  be	
  
combined	
  to	
  <ind:	
  
      •   The	
  TLDs	
  with	
  the	
  highest	
  ratio	
  of	
  transactions	
  to	
  domains	
  –	
  indicating	
  a	
  
          large	
  number	
  of	
  transactions	
  across	
  a	
  small	
  subset	
  of	
  domains.	
  	
  In	
  other	
  
          words,	
  there	
  are	
  only	
  a	
  few	
  unique	
  domains	
  in	
  the	
  TLD	
  that	
  make	
  it	
  popular.
      •   The	
  TLDs	
  with	
  the	
  lowest	
  ratio	
  of	
  transactions	
  to	
  domains	
  –	
  indicating	
  a	
  
          number	
  of	
  domains	
  among	
  which	
  the	
  transactions	
  are	
  spread	
  out.	
  In	
  other	
  
          words,	
  the	
  unique	
  domains	
  that	
  have	
  a	
  small	
  number	
  of	
  visits	
  or	
  transactions.	
  
October                                November                           December
Rank     TLD      Ratio                         TLD      Ratio                     TLD      Ratio
1    nu         5063                        nu         8083                    net        3737
2    net        3617                        net        3428                    nu         2824
3    ly         2140                        ly         1792                    ly         1699
4    tv         1719                        tv         1568                    tv         1692
5    pe         1326                        id         1267                    fm         1307
6    fm         1140                        pe         1159                    lan        1260
7    in         803                         lan        1153                    pe         1228
8    com        726                         fm         968                     com        799
9    it         707                         ir         807                     in         765
10   id         677                         com        702                     im         713
11   aero       676                         it         678                     pf         701
12   hn         662                         in         655                     it         633
13   tr         655                         th         622                     gov        592
14   au         520                         au         584                     th         587
15   su         500                         tr         531                     vn         546
16   im         483                         gr         515                     au         517
17   gov        480                         dk         503                     ir         485
18   ke         422                         local      471                     za         463
19   ph         409                         hn         398                     hn         461
20   ec         400                         gov        392                     tr         436
21   int        391                         mx         389                     mx         417
22   co         355                         ke         374                     sg         382
23   fr         341                         io         368                     ke         380
24   th         330                         co         331                     va         372
25   mx         315                         ec         325                     ec         368

Well	
  utilized,	
  generic	
  TLDs	
  (gTLD),	
  such	
  as	
  .com,	
  will	
  have	
  a	
  high	
  ratio	
  because	
  
domains	
  like	
  Google,	
  Facebook,	
  Amazon,	
  Yahoo,	
  Microsoft,	
  MySpace,	
  Twitter,	
  etc.	
  
contain	
  a	
  large	
  number	
  of	
  the	
  transactions	
  to	
  that	
  TLD.	
  This	
  is	
  however	
  offset	
  to	
  a	
  
certain	
  extent	
  because	
  there	
  are	
  also	
  a	
  large	
  number	
  of	
  popular	
  domains	
  on	
  these	
  
gTLDs	
  and	
  these	
  unique	
  domains	
  will	
  lower	
  the	
  ratio	
  somewhat,	
  though	
  it	
  remains	
  
relatively	
  high	
  overall.	
  For	
  example,	
  October	
  –	
  December	
  2009	
  saw	
  .com	
  ratios	
  of	
  
726:1,	
  702:1,	
  and	
  799:1	
  respectively.

It	
  is	
  interesting	
  to	
  further	
  analyze	
  domain	
  results	
  for	
  less	
  popular	
  TLDs	
  and	
  those	
  
that	
  had	
  a	
  higher	
  ratio	
  than	
  the	
  gTLDs,	
  both	
  from	
  a	
  statistical	
  and	
  trending	
  
perspective	
  as	
  well	
  as	
  from	
  a	
  security	
  perspective.	
  Miscreants	
  frequently	
  register	
  
domains	
  with	
  TLDs	
  that	
  are	
  less	
  in	
  demand	
  because	
  they	
  are	
  cheaper,	
  and	
  in	
  some	
  
cases	
  the	
  particular	
  domain	
  registry	
  (maintainer	
  of	
  the	
  TLD)	
  and/or	
  registrar	
  
(maintainer	
  of	
  the	
  domain	
  record)	
  will	
  have	
  poor	
  abuse	
  handling	
  procedures.	
  
Additionally,	
  the	
  registry	
  and/or	
  registrar	
  may	
  either	
  be	
  complicit	
  in	
  the	
  illegal	
  
activity	
  or	
  be	
  in	
  a	
  jurisdiction/country	
  with	
  a	
  legal	
  system	
  that	
  protects	
  the	
  domain	
  
from	
  being	
  de-­‐registered	
  or	
  having	
  the	
  registration	
  information	
  shared	
  with	
  law	
  
enforcement.	
  TLDs	
  with	
  a	
  high-­‐ratio	
  of	
  transactions	
  per	
  unique	
  domain	
  per	
  TLD	
  
have	
  one	
  or	
  more	
  domains	
  with	
  a	
  large	
  number	
  of	
  transactions.	
  It	
  is	
  interesting	
  to	
  
sift	
  through	
  the	
  records	
  to	
  explain	
  the	
  high-­‐ratio	
  TLDs.	
  They	
  may	
  be	
  the	
  result	
  of	
  a	
  
malicious	
  command	
  and	
  control	
  (C&C)	
  or	
  information	
  drop	
  server	
  that	
  has	
  a	
  large	
  
number	
  of	
  transactions	
  beaconing	
  to	
  the	
  domain’s	
  server,	
  or	
  it	
  could	
  be	
  something	
  
benign,	
  such	
  as	
  a	
  popular	
  social	
  networking	
  site	
  in	
  a	
  particular	
  country.

One	
  such	
  example	
  of	
  a	
  benign	
  domain	
  within	
  a	
  TLD	
  that	
  bubbled	
  to	
  the	
  top	
  was	
  .ly.	
  	
  
This	
  domain	
  had	
  a	
  ratio	
  of	
  2140:1,	
  1792:1,	
  and	
  1699:1	
  in	
  the	
  October	
  –	
  December	
  
timeframe.	
  These	
  ratios	
  were	
  more	
  than	
  double	
  the	
  ratios	
  that	
  .com	
  had	
  during	
  
these	
  months.	
  This	
  high	
  ratio	
  is	
  explained	
  by	
  this	
  TLD	
  being	
  relatively	
  unpopular	
  as	
  
far	
  as	
  unique	
  domains	
  go,	
  but	
  having	
  a	
  large	
  number	
  of	
  transactions	
  to	
  a	
  popular	
  
domain	
  -­‐	
  namely	
  bit.ly,	
  a	
  popular	
  URL	
  shortening	
  service.

The	
  .nu	
  TLD	
  had	
  even	
  higher	
  ratios	
  of	
  5063:1,	
  8083:1,	
  and	
  2824:1	
  in	
  Q4	
  2009.	
  
The	
  .nu	
  TLD	
  is	
  assigned	
  to	
  the	
  island	
  state	
  of	
  Niue,	
  and	
  Wikipedia	
  states	
  that	
  the	
  TLD	
  
“is	
  particularly	
  popular	
  in	
  Sweden,	
  Denmark,	
  the	
  Netherlands	
  and	
  Belgium,	
  as	
  nu	
  is	
  
the	
  word	
  for	
  ‘now’	
  in	
  Swedish,	
  Danish,	
  and	
  Dutch.”	
  While	
  the	
  domain	
  may	
  be	
  popular	
  
for	
  these	
  countries,	
  our	
  ratio	
  shows	
  that	
  a	
  relatively	
  small	
  number	
  of	
  domains	
  are	
  
dominating	
  the	
  transactions	
  for	
  this	
  TLD.

Running	
  a	
  query	
  against	
  the	
  Zscaler	
  NanoLogs	
  for	
  the	
  .nu	
  domains	
  and	
  count	
  of	
  
transactions,	
  yielded	
  a	
  large	
  percentage	
  of	
  the	
  transactions	
  to	
  the	
  domain:	
  
cvnxus.mine.nu.	
  The	
  transactions	
  to	
  the	
  domain	
  appear	
  as:

	
           hxxp://	
  cvnxus.mine.nu:53/30080000

Further	
  analysis	
  revealed	
  that	
  there	
  were	
  several	
  bot	
  infected	
  hosts	
  that	
  were	
  
beaconing	
  TCP	
  ACK	
  packets	
  to	
  this	
  host.	
  	
  Zscaler	
  has	
  since	
  noti<ied	
  and	
  assisted	
  
impacted	
  customers.	
  A	
  separate	
  white	
  paper	
  detailing	
  this	
  analysis	
  will	
  be	
  released.

Top	
  Domains	
  Visited
Many	
  of	
  the	
  most	
  visited	
  domains	
  are	
  actually	
  those	
  that	
  operate	
  behind	
  the	
  scenes.	
  
liveperson.net	
  for	
  example	
  is	
  a	
  real-­‐time	
  support	
  tool	
  used	
  by	
  a	
  variety	
  of	
  large	
  
online	
  retail	
  and	
  services	
  companies	
  such	
  as	
  Bank	
  of	
  America,	
  AT&T	
  and	
  IBM3.	
  As	
  
such,	
  when	
  receiving	
  email	
  and	
  chat	
  based	
  customer	
  support	
  at	
  such	
  companies	
  
certain	
  traf<ic	
  is	
  actually	
  redirected	
  to	
  the	
  liverperson.net	
  domain.	
  Top	
  domains	
  are	
  
calculated	
  based	
  on	
  the	
  total	
  number	
  of	
  transactions.	
  As	
  such,	
  sites	
  delivering	
  
images,	
  streaming	
  content	
  or	
  requiring	
  frequent	
  communication	
  of	
  some	
  form	
  tend	
  
to	
  score	
  higher.	
  Advertising	
  based	
  traf<ic	
  was	
  very	
  prevalent	
  with	
  ad	
  management	
  
platforms	
  such	
  as	
  doubleclick.net	
  and	
  yieldmanager.com,	
  both	
  landing	
  in	
  the	
  top	
  10.	
  
Google,	
  Yahoo!	
  and	
  Facebook	
  all	
  ranked	
  high,	
  as	
  did	
  domains	
  owned	
  and	
  managed	
  
by	
  them.	
  <bcdn.net	
  and	
  yimg.com	
  serve	
  up	
  Facebook	
  and	
  Yahoo!	
  content	
  
respectively.	
  google-­‐analytics.com,	
  a	
  Google	
  tool	
  for	
  tracking	
  site	
  visitors	
  receives	
  
signi<icant	
  traf<ic	
  due	
  to	
  the	
  fact	
  that	
  links	
  to	
  the	
  domain	
  are	
  posted	
  on	
  numerous	
  
third	
  party	
  sites.


3      http://solutions.liveperson.com/company/customers/
Top 10 Domains Visited By Month Q4 2009




                                               0%             7.50%             15.00%               22.50%                 30.00%

       liveperson.net

          google.com

      doubleclick.net
               fbcdn.net
              yahoo.com
                yimg.com
          facebook.com
google-analytics.com
      yieldmanager.com
              login.icq.com




                           October                            November                             December


CIDR	
  Block	
  Distribu0on
CIDR	
  notation	
  is	
  a	
  way	
  of	
  writing	
  a	
  block	
  of	
  IP	
  addresses,	
  where	
  the	
  suf<ix	
  number	
  is	
  
the	
  number	
  of	
  bits	
  to	
  include	
  from	
  the	
  IP	
  for	
  the	
  block4 .	
  For	
  example:

• 192.168.1.0/24	
  is	
  the	
  IP	
  block:	
  192.168.1.0-­‐192.168.1.255
• 192.168.0.0/16	
  is	
  the	
  IP	
  block:	
  192.168.0.0-­‐192.168.255.255
• 192.0.0.0/8	
  is	
  the	
  IP	
  block:	
  192.0.0.0-­‐192.255.255.255
The	
  chart	
  below	
  shows	
  the	
  top	
  25	
  most	
  popular,	
  highly	
  utilized	
  IP	
  blocks	
  based	
  on	
  
Zscaler	
  customer	
  traf<ic.	
  These	
  results	
  are	
  displayed	
  in	
  three	
  ways:	
  (1)	
  a	
  narrow,	
  /24	
  
IP	
  block,	
  viewpoint,	
  (2)	
  a	
  middle,	
  /16	
  IP	
  block,	
  viewpoint,	
  and	
  (3)	
  a	
  broader,	
  /8	
  IP	
  
block,	
  viewpoint.

The	
  narrow,	
  /24	
  IP	
  block,	
  viewpoint	
  is	
  largely	
  comprised	
  of	
  popular	
  end-­‐user	
  sites/
services	
  that	
  are	
  distributed	
  across	
  their	
  IP	
  block.	
  The	
  4th	
  quarter	
  included	
  some	
  of	
  

4   http://en.wikipedia.org/wiki/CIDR_notation
the	
  busiest	
  shopping	
  months	
  of	
  the	
  year.	
  This,	
  combined	
  with	
  Amazon's	
  utilization	
  
of	
  their	
  IP	
  blocks	
  (e.g.,	
  their	
  EC2	
  service),	
  accounted	
  for	
  Amazon	
  having	
  the	
  top	
  10	
  /
24	
  IP	
  blocks	
  by	
  number	
  of	
  unique	
  IPs	
  visited.	
  MySpace	
  and	
  Vkontakte	
  are	
  social	
  
networking	
  sites	
  that	
  seem	
  to	
  distribute	
  their	
  user	
  load	
  and/or	
  content	
  among	
  a	
  
number	
  of	
  web	
  server	
  IPs	
  in	
  their	
  block.

The	
  middle	
  /16	
  IP	
  block,	
  displays	
  some	
  of	
  the	
  more	
  popular	
  hosting	
  and	
  service	
  
providers	
  by	
  unique	
  IPs	
  visited,	
  such	
  as,	
  1&1,	
  Digital	
  United,	
  Taiwan	
  Fixed	
  Network,	
  
and	
  HiNet.	
  	
  It	
  is	
  interesting	
  that	
  when	
  looking	
  at	
  the	
  most	
  popular	
  IP	
  blocks	
  from	
  a	
  
middle	
  aggregation	
  point,	
  /16	
  IP	
  blocks,	
  more	
  Asia	
  based	
  IP	
  blocks	
  bubble	
  to	
  the	
  
top.	
  From	
  smaller	
  (/24	
  IP	
  blocks)	
  and	
  larger	
  (/8	
  IP	
  blocks)	
  IP	
  aggregation	
  points,	
  
more	
  United	
  States	
  based,	
  ARIN	
  space	
  <inds	
  its	
  way	
  into	
  the	
  top	
  25	
  blocks	
  by	
  unique	
  
IP	
  visited.	
  This	
  suggests	
  that	
  Asian	
  /	
  APNIC	
  service	
  and	
  hosting	
  providers	
  may	
  
largely	
  be	
  constructed	
  of	
  /16	
  or	
  similar	
  sized	
  blocks.	
  

                        /24 CIDR Block                                    /16 CIDR Block                           /8 CIDR Block
  Rank               Range      Organization                          Range       Organization                         Range
1             216.137.37.0/24         Amazon                    74.208.0.0/16          1&1 Internet Inc.            74.208.0.0/8
2             216.137.39.0/24         Amazon                    123.204.0.0/16         Digital United               69.0.0.0/8
3             216.137.41.0/24         Amazon                    124.8.0.0/16           Taiwan Fixed Network 216.0.0.0/8
4             216.137.45.0/24         Amazon                    114.44.0.0/16          HiNet                        66.0.0.0/8
5             216.137.47.0/24         Amazon                    219.85.0.0/16          Sony Network Taiwan 74.0.0.0/8
6             216.137.55.0/24         Amazon                    124.218.0.0/16         Asia Pacific On-line          208.0.0.0/8
7             216.137.59.0/24         Amazon                    122.121.0.0/16         HiNet                        64.0.0.0/8
8             216.137.53.0/24         Amazon                    220.136.0.0/16         HiNet                        72.0.0.0/8
9             216.137.43.0/24         Amazon                    125.230.0.0/16         HiNet                        67.0.0.0/8
10            216.137.61.0/24         Amazon                    114.47.0.0/16          HiNet                        61.0.0.0/8
11            63.135.88.0/24          MySpace                   112.104.0.0/16         Digital United               218.0.0.0/8
12            216.137.35.0/24         Amazon                    59.117.0.0/16          HiNet                        209.0.0.0/8
13            91.192.55.0/24          spamfighter.com            118.160.0.0/16         HiNet                        118.0.0.0/8
14            93.186.229.0/24         Vkontakte.ru              74.125.0.0/16          Google Inc.                  174.0.0.0/8
15            70.35.16.0/24           Netfirms, Inc.             218.172.0.0/16         HiNet                        65.0.0.0/8
16            93.186.230.0/24         Vkontakte.ru              69.192.0.0/16          Akamai Technologies 122.0.0.0/8
17            64.71.33.0/24           affinity.com               96.17.0.0/16           Akamai Technologies 207.0.0.0/8
18            69.89.31.0/24           bluehost.com              96.6.0.0/16            Akamai Technologies 87.0.0.0/8
19            65.54.81.0/24           Microsoft.com             118.171.0.0/16         HiNet                        220.0.0.0/8
20            64.12.24.0/24           aol.net                   219.81.0.0/16          Taiwan Fixed Network 124.0.0.0/8
21            124.218.196.0/24 Asia Pacific On-line 114.40.0.0/16                       HiNet                        125.0.0.0/8
22            124.218.194.0/24 Asia Pacific On-line 219.84.0.0/16                       Sony Network Taiwan 219.0.0.0/8
23            124.218.198.0/24 Asia Pacific On-line 218.163.0.0/16                      HiNet                        59.0.0.0/8
24            124.218.200.0/24 Asia Pacific On-line 61.31.0.0/16                        Taiwan Fixed Network 96.0.0.0/8
25            124.218.202.0/24 Asia Pacific On-line 114.43.0.0/16                       HiNet                        82.0.0.0/8

To	
  get	
  a	
  clearer	
  picture	
  of	
  actual	
  organizations	
  with	
  a	
  large	
  number	
  of	
  visited	
  web	
  
servers	
  (unique	
  web	
  server	
  IPs),	
  a	
  chart	
  was	
  created	
  breaking	
  out	
  unique	
  IPs	
  visited	
  
per	
  autonomous	
  system.	
  An	
  autonomous	
  system	
  (AS)	
  is	
  a	
  collection	
  of	
  connected	
  IP	
  
blocks	
  under	
  the	
  control	
  a	
  group/organization.	
  	
  The	
  <irst	
  and	
  third	
  most	
  popular	
  ASs	
  
are	
  Asian,	
  which	
  correlates	
  with	
  our	
  previous	
  statement.

ASN	
  Distribu0on
     Rank          ASN                               Organization                       Percentage
1                AS3462       HINET Data Communication Business Group                       13.36%
2                AS21844      ThePlanet.com Internet Services, Inc.                          2.31%
3                AS9924       Taiwan Fixed Network, Telco and Network Service Provider.      1.53%
4                AS2914       NTT America, Inc.                                              1.36%
5                AS8560       1&1 Internet AG                                                1.33%
6                AS7132       AT&T Internet Services                                         1.27%
7                AS4780       Digital United Inc.                                            1.21%
8                AS33070      Rackspace.com, Ltd.                                            1.02%
9                AS4134       CHINANET-BACKBONE No.31,Jin-rong Street                        1.01%
10               AS18182      Sony Network Taiwan Limited                                    1.01%
11               AS36351      SoftLayer Technologies Inc.                                    0.97%
12               AS26347      New Dream Network, LLC                                         0.95%
13               AS26496      GoDaddy.com, Inc.                                              0.95%
14               AS3269       TELECOM ITALIA                                                 0.86%
15               AS209        Qwest Communications Company, LLC                              0.82%
16               AS3356       Level 3 Communications                                         0.63%
17               AS20940      Akamai Technologies European AS                                0.62%
18               AS15169      Google Inc.                                                    0.53%
19               AS16276      OVH                                                            0.50%
20               AS32244      Liquid Web, Inc.                                               0.48%
21               AS3215       France Telecom - Orange                                        0.43%
22               AS12322      PROXAD AS for Proxad/Free ISP                                  0.41%
23               AS3549       Global Crossing Ltd.                                           0.33%
24               AS22822      Limelight Networks, Inc.                                       0.33%
25               AS7482       Asia Pacific On-line Service Inc.                               0.27%
Geography
The	
  majority	
  of	
  requested	
  web	
  content	
  resides	
  on	
  servers	
  located	
  in	
  the	
  United	
  
States.	
  With	
  the	
  exception	
  of	
  a	
  spike	
  in	
  October	
  and	
  part	
  of	
  November	
  for	
  content	
  
located	
  in	
  Taiwan,	
  traf<ic	
  not	
  destined	
  for	
  non-­‐US	
  based	
  web	
  sites	
  was	
  fairly	
  evenly	
  
distributed	
  across	
  servers	
  located	
  primarily	
  in	
  a	
  variety	
  of	
  countries	
  in	
  Europe	
  and	
  
Asia.
                                                   Top 10 Destinations By Country Q4 2009




                                   0%               15.00%              30.00%               45.00%                 60.00%


       United States
               Taiwan
             Germany
                France
    United Kingdom
              China
             Canada
                Italy
   Russian Federation
               Japan




                          October                          November                            December

       Country                        October             November              December
United States                            35.73%               44.42%                55.74%
Taiwan                                   34.22%               10.41%                 3.15%
Germany                                   2.76%                4.36%                 4.32%
France                                    1.82%                5.27%                 2.84%
United Kingdom                            2.21%                3.42%                 3.83%
China                                     2.01%                2.70%                 2.66%
Canada                                    1.97%                2.54%                 2.71%
Italy                                     2.55%                1.63%                 0.87%
Russian Federation                        1.42%                2.09%                 1.83%
Japan                                     1.64%                1.65%                 1.61%
Top 10 Destinations By Region Q4 2009



                                   0%             10.00%              20.00%                30.00%                  40.00%

           Taipei
      California
            Texas
Massachusetts
   Pennsylvania
          Arizona
         New York
              Illinois
              Taiwan
              Florida




                       October                          November                            December


            Region                      October            November              December
Taipei                                     30.14%               9.10%                2.67%
California                                  6.94%               7.23%                8.87%
Texas                                       4.87%               6.40%                8.33%
Massachusetts                               1.98%               2.71%                3.61%
Pennsylvania                                1.55%               2.06%                2.62%
Arizona                                     1.53%               2.00%                2.65%
New York                                    1.41%               1.92%                2.17%
Illinois                                    1.30%               1.71%                2.15%
Taiwan                                      3.26%               1.04%
Florida                                     1.25%               1.64%                    2.08%

From	
  a	
  regional	
  perspective,	
  Taipei	
  hosted	
  much	
  of	
  the	
  Taiwanese	
  based	
  content	
  
which	
  accounted	
  for	
  the	
  surge	
  in	
  October	
  and	
  November.	
  As	
  for	
  US	
  traf<ic,	
  both	
  
California	
  and	
  Texas	
  account	
  for	
  the	
  bulk	
  of	
  content,	
  with	
  the	
  remainder	
  tending	
  to	
  
be	
  located	
  on	
  the	
  East	
  Coast.
Top 10 Destinations By City Q4 2009


                  0%         10.00%       20.00%
                                                      30.00%
                                                                  40.00%
     Taipei
   Houston
 Cambridge
San Antonio
      Dallas
  Scottsdale
  Englewood
      Seattle
    Moscow
         Brea




                October           November             December


     City        October     November     December
Taipei              30.14%        9.10%       2.67%
Houston              1.87%        2.48%       3.21%
Cambridge            1.41%        1.98%       2.79%
San Antonio          1.08%        1.40%       1.85%
Dallas               0.98%        1.30%       1.73%
Scottsdale           0.75%        1.00%       1.34%
Englewood            0.77%        0.97%       1.30%
Seattle              0.71%        0.91%       1.17%
Moscow               0.71%        0.99%       0.93%
Brea                 0.66%        0.83%       1.12%
File	
  Types


                                                                         Many	
  assume	
  that	
  web	
  traf<ic	
  is	
  
                       Top 10 File Types Q4 2009                         dominated	
  by	
  HTML	
  content.	
  While	
  
                                                                         that	
  may	
  have	
  been	
  true	
  a	
  decade	
  
                                                                         ago,	
  the	
  media	
  rich,	
  dynamic	
  web	
  
                  0%     7.50% 15.00% 22.50%                             applications	
  available	
  today	
  are	
  
                                                           30.00%
                                                                         <illed	
  with	
  images,	
  formatting	
  
jpeg     28.74%                                                          elements,	
  data	
  and	
  active	
  content.	
  
  gif 28.68%                                                             For	
  the	
  4th	
  quarter	
  JPEG	
  (28.74%)	
  
   gz     24.40%                                                         and	
  GIF	
  (28.68%)	
  images	
  alone	
  
  png 6.25%                                                              accounted	
  for	
  more	
  than	
  half	
  of	
  the	
  
     js    3.95%                                                         total	
  number	
  of	
  transactions.	
  This	
  is	
  
   css 1.82%                                                             a	
  testament	
  to	
  the	
  visual	
  nature	
  of	
  
   swf 1.14%
                                                                         the	
  web.	
  JavaScript,	
  the	
  ‘work	
  horse’	
  
    xml 0.72%
       txt 0.66%                                                         of	
  modern,	
  user-­‐friendly	
  web	
  
      jpg 0.57%                                                          applications	
  was	
  responsible	
  for	
  
                                                                         only	
  3.95%	
  of	
  transactions.	
  HTML	
  
                                                                         <iles	
  fell	
  just	
  outside	
  of	
  the	
  top	
  10	
  
                                                                         and	
  drove	
  0.57%	
  of	
  traf<ic.

                                                                       Request	
  Method
                                                                        Predictably,	
  GET	
  requests	
  account	
  
  INVALID
                  0.04%
                  0.24%
                                                                        for	
  the	
  majority	
  of	
  traf<ic.	
  Generally	
  
                  0%                                                    speaking,	
  GET	
  and	
  POST	
  requests	
  
                  86.58%                                                are	
  the	
  most	
  often-­‐used	
  
         GET      83.46%
                  96.29%                                                communication	
  methods	
  employed	
  
                  7.18%                                                 by	
  web	
  applications,	
  the	
  difference	
  
    POST          15.16%                                                being	
  that	
  a	
  GET	
  request	
  passes	
  
                  3.67%
                  0.14%
                                                                        request	
  variables	
  within	
  the	
  URL	
  
    HEAD          0.07%                                                 itself	
  while	
  POST	
  requests	
  pass	
  
                  0%
                                                                        variables	
  as	
  a	
  portion	
  of	
  the	
  request	
  
                  0%
    MOVE          0%                                                    header.	
  Each	
  approach	
  has	
  
                  0%
                                                                        advantages	
  and	
  disadvantages	
  but	
  
                  6.04%
CONNECT           1.05%                                                 given	
  size	
  limitations	
  for	
  GET	
  
                  0.02%                                                 requests,	
  the	
  POST	
  method	
  tends	
  to	
  
              0%      25.00% 50.00% 75.00% 100.00%                      be	
  reserved	
  for	
  situations	
  such	
  as	
  
                                                                        <illing	
  out	
  a	
  web	
  form	
  when	
  more	
  
                                                                        substantial	
  amounts	
  of	
  data	
  need	
  to	
  
             Transactions                     Request Size
             Response Size
                                                                        be	
  transmitted,	
  while	
  GET	
  requests	
  
                                                                        are	
  leveraged	
  for	
  general	
  web	
  page	
  
                                                                        rendering.	
  It	
  is,	
  however,	
  
interesting	
  to	
  note	
  the	
  percentages	
  of	
  overall	
  traf<ic	
  related	
  to	
  request	
  and	
  response	
  
size.	
  Even	
  though	
  POST	
  requests	
  accounted	
  for	
  7.18%	
  of	
  transactions	
  during	
  the	
  
quarter,	
  (given	
  their	
  use	
  in	
  uploading	
  content),	
  such	
  requests	
  were	
  responsible	
  for	
  
more	
  than	
  twice	
  as	
  much	
  (15.16%)	
  of	
  total	
  outbound	
  web	
  traf<ic.	
  HTTP	
  CONNECT	
  
requests	
  by	
  contrast	
  have	
  the	
  opposite	
  effect.	
  Such	
  requests	
  are	
  used	
  to	
  initiate	
  
traf<ic	
  on	
  an	
  alternate	
  port.	
  As	
  such	
  the	
  requests	
  are	
  limited	
  in	
  size.	
  

Response	
  Code

                                                       Top 10 Response Codes Q4 2009


                                                0%         20.00%         40.00%          60.00%            80.00%

                       200 - OK
             304 - Not Modified                    8.94%                                                           78.98%
                    302 - Found                3.22%
       307 - Temporary Redirect                2.23%
                            Invalid            2.00%
                 404 - Not Found               1.35%
                204 - No Content               0.98%
                  403 - Forbidden              0.80%
        301 - Moved Permanently                0.60%
              206 - Partial Content            0.28%




Just	
  over	
  4	
  out	
  of	
  5	
  requests	
  (80.29%)	
  returned	
  a	
  200	
  level	
  (success)	
  code	
  
representing	
  that	
  the	
  content	
  had	
  been	
  delivered	
  and	
  no	
  further	
  action	
  was	
  
required.	
  300	
  level	
  (redirection)	
  codes,	
  indicating	
  	
  additional	
  action	
  required	
  by	
  the	
  
requesting	
  browser	
  overall	
  accounted	
  for	
  15%	
  of	
  traf<ic.	
  Client	
  errors	
  (400)	
  were	
  
relatively	
  rare	
  at	
  2.49%,	
  but	
  not	
  as	
  rare	
  as	
  server	
  errors	
  (500),	
  which	
  occurred	
  only	
  
0.11%	
  of	
  the	
  time.	
  
Web	
  Browser	
  Sta0s0cs
Browser	
  Version

                    Browser Market Share By Month Q4 2009




                                                                                                                         80.0%


                                                                                                                       60.0%


                                                                                                                      40.0%


                                                                                                                     20.0%
            IE
                      Firefox
                                    Safari                                                                          0%
                                                   Opera
                                                                 Chrome
                                                                                Unknown
                                                                                                   Other



                        October                          November                              December

While	
  Internet	
  Explorer	
  clearly	
  continues	
  to	
  dominate,	
  we	
  are	
  witnessing	
  a	
  slow	
  but	
  
steady	
  decline	
  in	
  overall	
  market	
  share.	
  Regardless,	
  other	
  browser	
  vendors	
  have	
  a	
  
long	
  way	
  to	
  go	
  before	
  they	
  will	
  surpass	
  the	
  long	
  standing	
  market	
  leader.	
  We	
  saw	
  a	
  
greater	
  than	
  6%	
  jump	
  in	
  market	
  share	
  for	
  Firefox,	
  during	
  the	
  month	
  of	
  December.	
  
However	
  this	
  can	
  be	
  largely	
  attributed	
  to	
  improved	
  detection	
  methods	
  as	
  opposed	
  to	
  
an	
  unexpected	
  surge	
  in	
  traf<ic.	
  You	
  will	
  note	
  that	
  Unknown	
  traf<ic	
  declined	
  a	
  similar	
  
amount	
  during	
  the	
  same	
  time	
  period.	
  Unknown	
  traf<ic	
  accounts	
  for	
  a	
  reasonable	
  
amount	
  of	
  traf<ic	
  as	
  today	
  -­‐	
  the	
  majority	
  of	
  desktop	
  applications	
  communicate	
  via	
  
HTTP/HTTPS	
  for	
  a	
  variety	
  of	
  reasons	
  including	
  the	
  retrieval	
  of	
  additional	
  content,	
  
providing	
  online	
  support,	
  downloading	
  patches	
  and	
  submitting	
  error	
  reports.	
  Safari,	
  
Opera	
  and	
  Chrome	
  combined,	
  continue	
  to	
  account	
  for	
  less	
  than	
  two	
  percent	
  of	
  the	
  
traf<ic	
  that	
  we’re	
  seeing.	
  It	
  will	
  be	
  interesting	
  to	
  watch	
  Chrome	
  in	
  the	
  coming	
  months	
  
as	
  Google	
  is	
  starting	
  to	
  leverage	
  its	
  reach	
  to	
  promote	
  the	
  browser.
Internet Explorer Breakdown Q4 2009                                        Looking	
  at	
  the	
  breakdown	
  of	
  
                                                                                 Internet	
  Explorer	
  traf<ic	
  for	
  the	
  
                                                                                 quarter	
  is	
  particularly	
  
                                                                                 concerning.	
  The	
  majority	
  of	
  
                          5%                                                     enterprises	
  continue	
  to	
  
                                                                                 maintain	
  Internet	
  Explorer	
  6.x	
  
                                                                                 as	
  their	
  browser	
  of	
  choice.	
  
                                1%                                               While	
  IE	
  6	
  continues	
  to	
  be	
  
                                                                                 supported	
  by	
  Microsoft,	
  
                                                                                 meaning	
  that	
  patches	
  are	
  
                                                                                 deployed	
  for	
  any	
  known	
  
                                                                                 vulnerabilities,	
  it	
  lacks	
  
                                                          48%                    numerous	
  security	
  features	
  
        46%
                                                                                 present	
  in	
  IE	
  7	
  and	
  8.	
  IE	
  6	
  does	
  
                                                                                 not	
  maintain	
  malicious	
  URL	
  and	
  
                                                                                 phishing	
  block	
  lists,	
  a	
  feature	
  
                                                                                 that	
  is	
  now	
  common	
  place	
  in	
  all	
  
                                                                                 major	
  browsers	
  and	
  is	
  even	
  
                                                                                 making	
  its	
  way	
  into	
  mobile	
  
                                                                                 browsers.	
  Additionally,	
  IE	
  6	
  
                                                                                 lacks	
  protections	
  such	
  as	
  Data	
  
                                                                                 Execution	
  Prevention	
  (DEP)	
  and	
  
              IE 6.0                 IE 7.0                   Address	
  Space	
  Layout	
  Randomization	
  
              IE 8.0                 IE Other                 (ASLR),	
  two	
  features	
  which	
  increase	
  the	
  
                                                              complexity	
  of	
  executing	
  shellcode	
  should	
  a	
  
remote	
  browser	
  exploit	
  be	
  uncovered.	
  During	
  the	
  Operation	
  Aurora	
  attack,	
  when	
  
Google,	
  Adobe	
  and	
  other	
  high	
  pro<ile	
  enterprises	
  were	
  allegedly	
  in<iltrated	
  by	
  
Chinese	
  attackers,	
  the	
  attacks	
  only	
  targeted	
  IE6.	
  While	
  IE	
  7	
  &	
  8	
  were	
  vulnerable	
  to	
  
the	
  same	
  attack	
  vector,	
  reliable	
  exploit	
  code	
  had	
  not	
  been	
  produced	
  for	
  these	
  
versions	
  on	
  the	
  browser	
  due	
  to	
  additional	
  protections	
  such	
  as	
  DEP	
  and	
  ASLR.	
  IE	
  8	
  
also	
  added	
  a	
  critical	
  feature	
  which	
  has	
  now	
  been	
  adopted	
  by	
  chrome	
  -­‐	
  the	
  inclusion	
  
of	
  cross-­‐site	
  scripting	
  protection,	
  yet	
  another	
  feature	
  that	
  IE	
  6	
  lacks.	
  It	
  is	
  vital	
  that	
  
enterprises	
  move	
  away	
  from	
  IE	
  6,	
  even	
  though	
  it	
  continues	
  to	
  be	
  supported	
  by	
  
Microsoft	
  and	
  adopt	
  IE	
  8	
  to	
  take	
  advantage	
  of	
  numerous	
  security	
  enhancements.	
  
Google	
  has	
  indirectly	
  taken	
  an	
  important	
  step	
  toward	
  forcing	
  this	
  change	
  by	
  
dropping	
  support	
  for	
  Google	
  Docs	
  and	
  Google	
  Sites	
  in	
  IE	
  6,	
  starting	
  in	
  March	
  2010.

                                                  October                      November                         December
Internet Explorer                                      75.31%                        73.77%                           72.21%
Firefox                                                 8.44%                         8.87%                           15.32%
Safari                                                  1.41%                         1.38%                            1.39%
Opera                                                   0.02%                         0.06%                            0.09%
Chrome                                                  0.06%                         0.02%                            0.03%
Unknown                                                13.04%                        14.18%                            9.33%
Other                                                   1.72%                         1.72%                            1.63%
                                                      100.00%                       100.00%                          100.00%
User	
  Sta0s0cs
Attackers	
  are	
  no	
  longer	
  targeting	
  web	
  and	
  email	
  servers.	
  Instead,	
  they	
  are	
  
focusing	
  on	
  the	
  weakest	
  link	
  in	
  the	
  security	
  chain	
  -­‐	
  end	
  users.	
  Whether	
  such	
  
attacks	
  leverage	
  technical	
  vulnerabilities,	
  or	
  more	
  likely,	
  social	
  engineering	
  
attacks,	
  web	
  based,	
  client-­‐side	
  attacks	
  are	
  the	
  most	
  common	
  way	
  to	
  
compromise	
  end	
  user	
  machines.	
  As	
  such	
  it’s	
  vital	
  for	
  enterprises	
  to	
  
understand	
  user	
  behavior	
  on	
  the	
  web.
URL	
  Categoriza0on
Given	
  the	
  corporate	
  focus	
  of	
  Zscaler	
  clients	
  it	
  isn’t	
  surprising	
  that	
  categories	
  such	
  as	
  
Professional	
  Services	
  and	
  Corporate	
  Marketing	
  would	
  top	
  the	
  list.	
  More	
  interesting	
  
are	
  the	
  high	
  placements	
  of	
  personal	
  traf<ic	
  such	
  as	
  Shopping,	
  Sports,	
  Entertainment	
  
and	
  games.	
  In	
  fact,	
  the	
  majority	
  of	
  sites	
  beyond	
  the	
  top	
  10	
  are	
  personal	
  in	
  nature.	
  
Overall,	
  approximately	
  1/5	
  of	
  traf<ic	
  could	
  be	
  deemed	
  to	
  be	
  personal	
  in	
  nature.	
  
While	
  Zscaler	
  delivers	
  an	
  enterprise	
  offering	
  it	
  is	
  not	
  uncommon	
  for	
  employees	
  to	
  
leverage	
  corporate	
  assets	
  after	
  work	
  hours	
  for	
  personal	
  purposes	
  and	
  as	
  such,	
  some	
  
of	
  this	
  traf<ic	
  was	
  likely	
  generated	
  outside	
  of	
  work	
  hours.

Below	
  we	
  breakdown	
  a	
  select	
  number	
  of	
  individual	
  categories	
  to	
  reveal	
  the	
  top	
  10	
  
domains	
  within	
  each.

Search	
  Engines
                                                        The	
  search	
  engine	
  game	
  remains	
  a	
  three	
  horse	
  
          Top Search Engines                            race,	
  with	
  Google	
  continuing	
  to	
  dominate	
  the	
  
                                                        majority	
  of	
  traf<ic.	
  After	
  the	
  big	
  three,	
  
                                                        contenders	
  are	
  hard	
  to	
  <ind.	
  Disney’s	
  Go.com	
  
           Other                                        which	
  is	
  actually	
  powered	
  by	
  Yahoo!,	
  sat	
  in	
  4th	
  
           15%                                          place	
  at	
  1.22%	
  and	
  Baidu,	
  a	
  powerhouse	
  in	
  the	
  
                                                        Chinese	
  market	
  handled	
  1.00%	
  of	
  web	
  search	
  
  Microsoft                                             traf<ic	
  for	
  Q4	
  2009.
    10%


    Yahoo!                            Google
     18%                               57%
Social	
  networking
                                                                           The	
  dominance	
  of	
  Facebook	
  in	
  
 Top Social Networking Sites Q4 2009                                       the	
  social	
  networking	
  realm	
  is	
  
                                                                           clear.	
  Three	
  quarters	
  of	
  all	
  social	
  
                                                                           networking	
  traf<ic	
  traversing	
  the	
  
                                                                           Zscaler	
  network	
  is	
  destined	
  for	
  
                   Other                                                   Facebook.	
  MySpace	
  has	
  solid	
  
                   11%                                                     control	
  of	
  second	
  place	
  with	
  
                                                                           15%	
  of	
  traf<ic	
  but	
  the	
  gap	
  
                                                                           between	
  <irst	
  and	
  second	
  place	
  
   Myspace
                                                                           is	
  enormous	
  and	
  only	
  appears	
  to	
  
    15%
                                                                           be	
  getting	
  larger.

                                                                      It	
  is	
  also	
  interesting	
  to	
  consider	
  
                                                                      that	
  the	
  majority	
  of	
  traf<ic	
  in	
  
                                                                      these	
  statistics	
  is	
  corporate	
  
                                                                      traf<ic.	
  While	
  a	
  portion	
  of	
  
                                                                      requests	
  are	
  no	
  doubt	
  personal	
  
                                          Facebook                    in	
  nature,	
  this	
  also	
  suggests	
  that	
  
                                             74%
                                                                      Facebook	
  is	
  becoming	
  a	
  social	
  
                                                                      platform	
  of	
  choice	
  for	
  
                                                                      enterprises.	
  More	
  and	
  more,	
  
                                                                      corporations	
  are	
  attempting	
  to	
  
leverage	
  social	
  networks	
  for	
  marketing,	
  recruiting	
  and	
  investigating	
  potential	
  new	
  
hires.

File	
  Sharing
                                                                            Si@mBIT,	
  a	
  Thailand	
  based	
  web	
  
                              Top File Sharing Domains                      hosting	
  provider	
  describing	
  
                                                                            itself	
  as	
  “the	
  best	
  Thailand	
  
                                                                            Bittorrent	
  website	
  since	
  2005”	
  
                            0% 10.00% 20.00% 30.00%                         led	
  statistics	
  for	
  the	
  quarter	
  
                                                    40.00%
                                                                            with	
  37.49%	
  of	
  all	
  <ile	
  sharing	
  
siambit.com       37.49%
                                                                            traf<ic.	
  The	
  third	
  largest	
  
                                                                            domain,	
  tb.in.th	
  is	
  also	
  
filestube.com       25.31%
                                                                            controlled	
  by	
  Si@mBIT	
  	
  and	
  
       tb.in.th    12.98%
                                                                            commanded	
  12.98%	
  of	
  traf<ic,	
  
    sftcdn.net      9.68%
                                                                            giving	
  Si@mBIT	
  approximately	
  
iptorrents.com      3.06%                                                   half	
  of	
  the	
  traf<ic	
  for	
  the	
  quarter.	
  
 limewire.com        2.95%                                                  FilesTube,	
  a	
  search	
  engine	
  
         Other       1.98%                                                  dedicated	
  to	
  <ile	
  downloads	
  had	
  
 seedpeer.com        1.07%
                                                                            25.31%	
  of	
  traf<ic.	
  
Government
                                                                 Q4	
  means	
  that	
  Christmas	
  is	
  on	
  
                           Top 10 Government Domains             the	
  way	
  and	
  it	
  would	
  appear	
  
                                                                 that	
  the	
  United	
  States	
  Postal	
  
                                                                 Service	
  (USPS)	
  was	
  a	
  popular	
  
                         0%     5.00% 10.00% 15.00%
                                                                 destination	
  for	
  holiday	
  
                                                       20.00%
                                                                 shoppers	
  looking	
  to	
  determine	
  
                                                                 if	
  their	
  gifts	
  would	
  arrive	
  on	
  
 usps.com      16.20%
                                                                 time.	
  USPS	
  accounted	
  for	
  
  nraila.org     5.14%
                                                                 16.20%	
  of	
  government	
  related	
  
weather.gov      3.60%
                                                                 traf<ic	
  during	
  the	
  quarter.	
  
  uspto.gov      2.25%

www.sec.gov      1.73%

   state.fl.us     1.47%

    fema.gov      1.38%
 www.irs.gov       1.36%
 michigan.gov      1.12%
  military.com     1.03%




Retail
                                                                  Q4	
  is	
  of	
  course	
  the	
  peak	
  online	
  
                                 Top 10 Shopping Sites            shopping	
  season,	
  a	
  time	
  when	
  
                                                                  retailers	
  look	
  to	
  make	
  the	
  
                                                                  majority	
  of	
  their	
  pro<it	
  for	
  the	
  
                           0%    1.00% 2.00%   3.00%     4.00%    year.	
  If	
  web	
  traf<ic	
  is	
  any	
  
                                                                  indication,	
  Amazon	
  was	
  the	
  
      Amazon 3.63%                                                big	
  winner,	
  having	
  claimed	
  
ShopLocal.com 2.90%                                               3.63%	
  of	
  total	
  retail	
  traf<ic.	
  
       Macy’s    2.59%
                                                                  ShopLocal,	
  which	
  took	
  the	
  
     Shop.com 2.55%
     Overstock 1.88%
                                                                  number	
  two	
  spot,	
  is	
  not	
  a	
  
      JC Penny 1.87%                                              retailer	
  itself	
  but	
  rather	
  a	
  site	
  
          Target 1.52%                                            which	
  republishes	
  <lyers	
  for	
  
         Costco 1.27%                                             local	
  stores	
  to	
  allow	
  user	
  to	
  
  Barnes & Noble 1.10%
            QVC 1.10%                                             <ind	
  deals	
  speci<ic	
  to	
  their	
  
                                                                  geographic	
  area.	
  The	
  company	
  
                                                                  makes	
  money	
  through	
  
                                                                  advertising	
  on	
  the	
  site.
Security	
  Sta0s0cs
Threats
Next,	
  we’ll	
  breakdown	
  the	
  various	
  threats	
  that	
  we	
  see	
  on	
  a	
  daily	
  basis.	
  These	
  results	
  
are	
  based	
  on	
  actual	
  end	
  user	
  traf<ic	
  and	
  therefore	
  re<lect	
  popular	
  and	
  active	
  
malicious	
  sites	
  as	
  opposed	
  to	
  sites	
  that	
  may	
  exist	
  but	
  not	
  be	
  visited.

Malware	
  By	
  IP	
  Address
                                                                              Worms,	
  viruses,	
  Trojans	
  and	
  
                             Top 10 Malware IP Addresses                      other	
  forms	
  of	
  malware	
  can	
  be	
  
                                                                              found	
  just	
  about	
  everywhere	
  on	
  
                                                                              the	
  web	
  today.	
  However	
  
                               0% 10.00%20.00% 30.00% 40.00%                  malicious	
  content	
  is	
  not	
  
                                                                              necessarily	
  hosted	
  at	
  sites	
  that	
  
38.99.186.14          38.63%                                                  are	
  themselves	
  malicious.	
  More	
  
208.71.120.24 25.68%                                                          and	
  more,	
  we’re	
  seeing	
  otherwise	
  
208.71.121.24 13.41%                                                          legitimate	
  sites	
  hosting	
  malware	
  
 124.153.77.48 5.52%                                                          without	
  them	
  being	
  aware	
  of	
  it.	
  
     217.23.7.7 2.73%                                                         This	
  is	
  an	
  increasing	
  concern	
  
    64.14.29.50 1.32%
                                                                              given	
  the	
  trend	
  toward	
  
216.86.150.237 1.01%
   208.76.70.56 1.00%                                                         permitting	
  user	
  supplied	
  content	
  
   74.125.19.83 0.91%                                                         to	
  be	
  shared.	
  Unfortunately,	
  many	
  
    74.125.19.18 0.88%                                                        sites	
  are	
  doing	
  little	
  to	
  ensure	
  
                                                                              that	
  the	
  hosted	
  content	
  is	
  not	
  
                                                                              malicious	
  before	
  it	
  is	
  stored	
  for	
  
                                                                              others	
  to	
  access.
Malware	
  by	
  Country
Sites	
  hosted	
  in	
  the	
  United	
  States	
  
overwhelmingly	
  hosted	
  the	
  majority	
  of	
                       Top Countries Serving Malware
malware	
  and	
  for	
  this	
  reason	
  we	
  have	
  
broken	
  them	
  out	
  separately.	
  80.32%	
  of	
  
malware	
  seen	
  during	
  Q4	
  2009	
  originated	
  
from	
  US	
  based	
  servers.	
  This	
  should	
  not,	
  
however,	
  be	
  interpreted	
  as	
  US-­‐based	
                               Other
traf<ic	
  being	
  particularly	
  risky,	
  rather	
  it’s	
                    20%
more	
  of	
  a	
  re<lection	
  of	
  the	
  fact	
  that	
  the	
  
majority	
  of	
  traf<ic	
  inspected	
  was	
  destined	
  
for	
  served	
  located	
  in	
  the	
  US.	
  This	
  can	
  be	
  
seen	
  in	
  the	
  Geography	
  section	
  of	
  this	
  
paper.
                                                                                                 United States
                                                                                                     80%
Top 10 Countries Serving Malware (US Excluded)




                                              5% 3%
                                     5%
                                                                                   25%
                               6%

                            6%

                             6%


                                   11%
                                                                                   20%
                                                     14%


          Netherlands                   India                      Germany            China                         Cyprus
          Russian Federation            United Kingdom             Canada             Korea, Republic of            France

Phishing
                                                                                  The	
  top	
  phishing	
  site	
  blocked	
  
                            Top 10 Phishing IP Addresses                          was	
  coolxd.com	
  -­‐	
  this	
  
                                                                                  accounted	
  for	
  roughly	
  70%	
  of	
  
                                                                                  the	
  quarter's	
  phishing	
  
                                                                     80.00% numbers.	
  The	
  site	
  itself,	
  was	
  
                             0% 20.00% 40.00% 60.00%
                                                                                  recently	
  removed	
  from	
  the	
  
208.43.210.147 70.83%                                                             Internet.	
  This	
  scam	
  site	
  is	
  
219.232.243.74 7.21%                                                              effectively	
  the	
  same	
  as	
  the	
  
 219.232.243.65 1.91%
                                                                                  heyxd.com,	
  omgxd.com,	
  and	
  
 219.232.243.91 1.64%
  219.232.243.75 1.47%                                                            imnotez.com	
  sites.	
  These	
  sites	
  
  219.232.243.15 0.84%                                                            steal	
  your	
  email/instant	
  
  219.232.243.90 0.61%                                                            messenger	
  credentials	
  
 219.232.241.178 0.57%
   219.232.243.87 0.55%                                                           (username/password),	
  and	
  
     174.143.29.2 0.50%                                                           then	
  noti<ies	
  the	
  people	
  on	
  
                                                                                  your	
  contact	
  list	
  to	
  check	
  out	
  
                                                                                  the	
  site.	
  Advertisements,	
  
                                                                                  fraud,	
  and/or	
  malware	
  are	
  
                                                                                  then	
  spammed	
  to	
  and	
  through	
  
victim	
  accounts.	
  The	
  sites	
  advertised	
  the	
  ability	
  to	
  provide	
  a	
  service	
  which	
  enables	
  
users	
  to	
  IM	
  pictures	
  and	
  other	
  content	
  to	
  share	
  directly	
  to	
  a	
  forum.
Malicious	
  Domains
                                                                                   Three	
  domains	
  accounted	
  for	
  
                                   Top 10 Malicious Domains                        roughly	
  55%	
  of	
  the	
  malicious	
  
                                                                                   URLs	
  transactions:

                                         0% 10.00% 20.00% 30.00%                   •adfarm.mediaplex.com
                                                                                   •link4you.3322.org
adfarm.mediaplex.com 24.01%                                                        •www.tns-­‐counter.ru
     link4you.3322.org 17.41%
    www.tns-counter.ru 13.33%                                                     adfarm.mediaplex.com,	
  has	
  been	
  
     www.winifixer.com 4.06%                                                       reported	
  to	
  be	
  involved	
  in	
  spam,	
  
    www.freegaming.de 2.96%                                                       adware/spyware,	
  phishing/
   dt.tongji.linezing.com 2.25%
         img.12chan.org 1.72%                                                     scams,	
  and	
  browser	
  exploits5.	
  
          nspmotion.com 1.14%                                                     The	
  Mediaplex	
  website	
  details	
  
                acs86.com  0.69%                                                  how	
  the	
  company	
  "provides	
  
    stork27.dropbox.com 0.66%
                                                                                  cross-­‐channel	
  advertising	
  
                                                                                  technology	
  solutions	
  and	
  
                                                                                  services	
  that	
  enable	
  marketers	
  to	
  
                                                                                                                achieve	
  one-­‐
    Top Malicious Domains By Country                                       United States                        to-­‐one	
  
                                                                           Canada                               messaging,	
  
                                                                           Russian Federation                   greater	
  
                                                                           China                                ef<iciencies	
  
                     6%                                                    Germany                              and	
  a	
  
           3%                                                              Netherlands
         5%                                                                                                     competitive	
  
                                                                           Other
                                                                                                                edge	
  through	
  
     6%                                                                                                         insightful	
  
                                                                        reporting	
  and	
  analytics”      6.	
  3322.org	
  is	
  a	
  

                                                                        DynDNS	
  provided	
  domain	
  that	
  has	
  
                                                       44%              served	
  malware	
  and	
  exploit	
  content	
  for	
  
    17%                                                                 some	
  time7.	
  tns-­‐counter.ru	
  is	
  also	
  
                                                                        known	
  for	
  serving	
  adware/spyware/
                                                                        malware 8.

                                                                 The	
  majority	
  of	
  malicious	
  sites	
  are	
  
                    19%                                          hosted	
  in	
  the	
  US,	
  with	
  a	
  full	
  63%	
  of	
  sites	
  
                                                                 residing	
  in	
  North	
  America.	
  This	
  is	
  
                                                                 however	
  more	
  a	
  re<lection	
  of	
  where	
  
content	
  in	
  general	
  resides	
  as	
  opposed	
  to	
  North	
  American	
  content	
  representing	
  a	
  
higher	
  overall	
  risk.

5   http://www.siteadvisor.com/sites/mediaplex.com/summary/
6   http://www.mediaplex.com/about.shtml
7   http://isc.sans.org/diary.html?storyid=5710

8   http://www.siteadvisor.com/sites/tns-­‐counter.ru/summary/
Anonymizers
                                           Top 10 Anonymizers
                                                                                       Over	
  30%	
  of	
  our	
  
                                                                                       anonymizer	
  traf<ic	
  was	
  to	
  
                                                                                       kproxy.com.	
  One	
  of	
  the	
  
                                                                                       features	
  that	
  Zscaler	
  
                                   0% 10.00% 20.00% 30.00%
                                                                             40.00% provides	
  to	
  customers,	
  is	
  
                                                                                       policy	
  based	
  blocking	
  based	
  
         kproxy.com 30.51%                                                             on	
  page	
  categorization.	
  So	
  
proxyswitcher.com 20.17%                                                               customers	
  have	
  the	
  ability	
  
   freeproxylist.org 8.03%                                                             to	
  block	
  users	
  from	
  
          archive.org       5.36%                                                      browsing	
  to/through	
  proxy	
  
          freeproxy.ru 3.12%                                                           sites.	
  kproxy.com	
  provides	
  
  privacy-world.com 1.83%
                                                                                       a	
  simple	
  interface,	
  not	
  
            helllabs.net 1.76%
                                                                                       unlike	
  Google’s,	
  to	
  browse	
  
         66.232.118.93 1.66%
      proxybridge.com 1.57%                                                            through,	
  with	
  SSL	
  
            ktunnel.com 1.39%                                                          encryption	
  as	
  an	
  additional	
  
                                                                                       capability.	
  Of	
  the	
  popular	
  
                                                                                       sites	
  that	
  kproxy	
  advertises	
  
                                                                                       that	
  it	
  works	
  with	
  are	
  
                                                                                       MySpace,	
  Facebook,	
  Gmail,	
  
YouTube,	
  and	
  MegaUpload	
  -­‐	
  all	
  sites,	
  that	
  may	
  be	
  blocked	
  by	
  company	
  policies	
  as	
  
they	
  are	
  not	
  work	
  related.	
  In	
  other	
  words,	
  users	
  are	
  generally	
  using	
  these	
  services	
  to	
  
get	
  around	
  corporate	
  policies	
  and	
  URL	
  <iltering	
  rules	
  as	
  opposed	
  to	
  using	
  them	
  to	
  
cloak	
  their	
  IP	
  address	
  from	
  an	
  external	
  source.

Botnets
                                                                                        Generally	
  speaking,	
  by	
  
                                  Top 10 Botnets IPs/Domains                            correlating	
  the	
  malicious	
  
                                                                                        artifact	
  to	
  the	
  top	
  botnet	
  
                                                                                        hosts,	
  enables	
  us	
  to	
  
                               0%    12.50% 25.00% 37.50%                               describe	
  which	
  malware	
  
                                                                          50.00%
                                                                                        campaigns	
  were	
  the	
  most	
  
 91.212.65.13        44.11%                                                             successful.	
  The	
  breakdown	
  
  66.235.175.5        15.67%                                                            is	
  as	
  follows,	
  and	
  should	
  
77.221.133.227        9.80%                                                             not	
  be	
  of	
  surprise	
  to	
  the	
  
    88.80.7.152        8.39%                                                            security	
  community	
  for	
  
       88.80.5.3       7.05%                                                            HTTP	
  based	
  botnets:
 77.221.133.189        5.74%
 208.99.193.130         3.18%                                                           1.Zeus/Zbot	
  variants
      meu89.net         1.91%                                                           2.Fake	
  Anti-­‐Virus	
  variants
    194.68.45.50         1.63%
                                                                                        3.Banker	
  Trojan	
  variants.
     69.61.21.115        0.28%

                                                                               The	
  top	
  command	
  and	
  
                                                                               control	
  IP	
  address	
  seen,	
  
                                                                               91.212.65.13,	
  is	
  based	
  out	
  
                                                                               of	
  the	
  Ukraine	
  and	
  serviced	
  
both	
  Zeus	
  and	
  FakeAV	
  infections.	
  The	
  whois	
  information	
  for	
  this	
  host	
  shows	
  it	
  
belonging	
  to	
  the	
  Eurohost/UralComp	
  IP	
  blocks.	
  FireEye	
  has	
  a	
  good	
  write-­‐up	
  of	
  this	
  
"bad	
  actor"	
  from	
  almost	
  a	
  year	
  ago9 	
  and	
  malwaredomainlist,	
  an	
  archive	
  of	
  malicious	
  
web	
  domains	
  has	
  plenty	
  of	
  content	
  for	
  these	
  IP	
  blocks10 .

While	
  Ukraine	
  and	
  Russian	
  IPs	
  make	
  up	
  a	
  large	
  number	
  of	
  the	
  botnet	
  C&C	
  servers,	
  it	
  
was	
  a	
  little	
  surprising	
  to	
  see	
  that	
  Sweden	
  had	
  a	
  number	
  of	
  C&Cs	
  in	
  the	
  top	
  25:
       •   88.80.7.152
       •   88.80.5.3
       •   88.80.5.172
       •   80.88.108.18

Further	
  analysis	
  of	
  some	
  of	
  the	
  Swedish	
  hosts	
  shows	
  them	
  belonging	
  to	
  PRQ	
  (http://
www.prq.se)	
  a	
  co-­‐location	
  and	
  hosting	
  provider.	
  	
  Their	
  homepage	
  states	
  that	
  they	
  
are	
  known	
  for	
  their	
  "boundless	
  commitment	
  to	
  free	
  speech"	
  and	
  "discrete	
  customer	
  
relations	
  policy".	
  They	
  also	
  have	
  an	
  icon	
  on	
  their	
  website	
  that	
  states,	
  "data	
  retention	
  
is	
  no	
  solution",	
  suggesting	
  minimal/no	
  logging.	
  	
  In	
  other	
  words,	
  this	
  hosting	
  service	
  
would	
  be	
  ideal	
  for	
  hosting	
  malicious	
  sites	
  and	
  remaining	
  protected	
  from	
  
investigations	
  /	
  takedowns.

Traffic
Last,	
  but	
  not	
  least,	
  we’ll	
  investigate	
  traf<ic	
  patterns	
  which	
  would	
  not	
  be	
  expected	
  
without	
  the	
  presence	
  of	
  errors	
  of	
  malicious	
  content.

Bogon	
  IP	
  space
                             Top 10 Bogon IP Addresses
                                                                              Bogon	
  (aka	
  darknet)	
  IP	
  addresses	
  
                                                                              represent	
  non-­‐routable	
  IP	
  blocks,	
  
                                                                              either	
  because	
  they	
  are	
  reserved	
  
                       0%      2.00%      4.00%        6.00%        8.00%     (for	
  example	
  RFC1918)	
  or	
  they	
  
                                                                              are	
  unallocated.	
  Occasionally,	
  we	
  
    1.1.1.1      7.74%
  127.0.0.0      6.60%                                                        see	
  web	
  requests	
  to	
  bogon	
  IPs	
  -­‐	
  
198.18.1.18       5.35%                                                       usually	
  this	
  is	
  to	
  RFC1918	
  
     1.2.3.4      4.34%
     0.0.0.2       2.99%                                                      address	
  (internal	
  IP	
  addresses),	
  
     0.0.0.5       2.62%
 198.18.1.15        2.60%                                                     and	
  the	
  requests	
  have	
  leaked	
  into	
  
      0.0.0.8       2.57%
                                                                              the	
  cloud	
  because	
  of	
  a	
  routing	
  
  198.18.1.2         2.20%
      0.0.0.1        2.20%                                                    miscon<iguration	
  on	
  the	
  




9   http://blog.fireeye.com/research/2009/03/bad-actors-part-6-eurohost-llc.html
10   http://www.malwaredomainlist.com/forums/index.php?board=23.0
Top 10 Bogon IP Address Blocks                        customer's	
  network.	
  However,	
  
                                                                                  there	
  are	
  also,	
  several	
  
                              0%
                                                                                  occurrences	
  of	
  web	
  requests	
  to	
  
                                      1.75% 3.50%          5.25%        7.00%     non-­‐RFC1918	
  bogons.	
  This	
  traf<ic	
  
   127.0.0.0/24         6.60%
                                                                                  is	
  of	
  interest	
  as	
  it	
  represents	
  
      1.2.3.0/24        4.34%                                                     either	
  human	
  error	
  or	
  an	
  infected	
  
      0.0.0.0/24         2.62%
     50.0.0.0/24         2.04%                                                    machine	
  that	
  is	
  randomly	
  
  169.254.1.0/24          1.11%                                                   scanning	
  IP	
  address	
  blocks	
  
169.254.178.0/24          0.56%
169.254.200.0/24           0.53%                                                  looking	
  for	
  vulnerable	
  hosts.
   169.254.8.0/24          0.37%
  198.18.189.0/24           0.37%
        0.1.0.0/24          0.34%                                                 Some	
  of	
  the	
  bogon	
  traf<ic	
  can	
  be	
  
                                                                                  explained	
  as	
  follows:



• The	
  1.1.1.1	
  and	
  127.0.0.0/8	
  and	
  1.2.3.0/24	
  subnets	
  are	
  likely	
  some	
  sort	
  of	
  test	
  
  scripts	
  that	
  folks	
  are	
  running.
• The	
  169.254.0.0/16	
  addresses	
  are	
  part	
  of	
  the	
  Automatic	
  Private	
  Addressing	
  
  (APIPA)	
  of	
  hosts	
  when	
  DHCP	
  fails.

The	
  50.0.0.0/24	
  IP	
  block	
  is	
  interesting,	
  though	
  yet	
  unexplained.	
  Googling	
  for	
  it	
  
shows	
  that	
  it	
  is	
  an	
  IANA	
  reserved	
  block,	
  and	
  it	
  shows	
  up	
  in	
  some	
  OSPF	
  routing	
  
templates.	
  It's	
  possible	
  that	
  this	
  block	
  is	
  a	
  commonly	
  used	
  reserve	
  block	
  in	
  some	
  
intra-­‐organization	
  routing.	
  However,	
  the	
  only	
  IP	
  address	
  that	
  was	
  hit	
  in	
  this	
  block	
  
was	
  50.0.0.82,	
  which	
  is	
  interesting.	
  It	
  is	
  possible	
  that	
  there	
  was	
  a	
  mistake	
  in	
  a	
  script	
  
or	
  routing	
  statement.
Conclusion
Understanding	
  web	
  traf<ic	
  is	
  critical	
  for	
  enterprises	
  seeking	
  to	
  manage	
  and	
  secure	
  
their	
  networks.	
  Traf<ic	
  is	
  converging	
  on	
  the	
  web	
  at	
  a	
  rapid	
  pace.	
  A	
  decade	
  ago	
  we	
  
leveraged	
  <irewalls	
  to	
  manage	
  traf<ic	
  on	
  networks	
  and	
  determine	
  which	
  users	
  could	
  
access	
  which	
  resources.	
  Today,	
  traf<ic	
  is	
  not	
  neatly	
  segregated	
  into	
  buckets	
  based	
  on	
  
protocols.	
  Regardless	
  of	
  the	
  traf<ic	
  that	
  we’re	
  dealing	
  with,	
  be	
  it	
  email,	
  instant	
  
messaging,	
  P2P,	
  streaming	
  media,	
  etc.,	
  it	
  has	
  the	
  ability	
  to	
  be	
  tunneled	
  through	
  
HTTP/HTTPS.

At	
  the	
  same	
  time,	
  attackers	
  have	
  shifted	
  their	
  focus	
  to	
  target	
  end	
  users.	
  Some	
  
attackers	
  take	
  a	
  shotgun	
  approach	
  by	
  striking	
  far	
  and	
  wide	
  without	
  concern	
  for	
  who	
  
the	
  ultimate	
  victims	
  may	
  be.	
  This	
  is	
  the	
  approach	
  leveraged	
  by	
  those	
  who	
  build	
  
botnets.	
  They	
  seek	
  infected	
  machines	
  and	
  they	
  do	
  not	
  discriminate.	
  On	
  the	
  other	
  
side	
  of	
  the	
  coin,	
  Advanced	
  Persistent	
  Threats11 	
  are	
  emerging	
  on	
  the	
  radars	
  of	
  CISOs	
  
as	
  the	
  media	
  highlights	
  the	
  sophistication	
  of	
  attacks	
  on	
  corporations,	
  such	
  as	
  those	
  
highlighted	
  in	
  the	
  Operation	
  Aurora	
  attacks	
  which	
  targeted	
  Google,	
  Adobe	
  and	
  
others.	
  Regardless	
  of	
  the	
  approach,	
  the	
  majority	
  of	
  such	
  attacks	
  now	
  leverage	
  the	
  
web	
  as	
  the	
  transport	
  medium.	
  

Understanding	
  the	
  behaviors	
  of	
  end	
  users,	
  content	
  providers	
  and	
  attackers	
  on	
  the	
  
web	
  can	
  help	
  us	
  to	
  better	
  manage	
  and	
  secure	
  networks.	
  We	
  hope	
  that	
  you	
  enjoyed	
  
this,	
  our	
  <irst	
  quarterly	
  State	
  of	
  the	
  Web	
  report.	
  




11   http://www.zscaler.com/apt.html
Appendix
TLD	
  Breakdown
Monthly	
  Summary	
  –	
  Top	
  TLDs	
  Visited
Note: Pink shows larger fluctuations than yellow, and green shows no fluctuation
               Monthly Summary – Top TLDs by Transactions
  Popularity     October 2009      November 2009       December 2009

      1              COM                COM                 COM
      2              NET                NET                 NET
      3              ORG                ORG                 ORG
      4               AU                 AU                 AU
      5               IN                 UK                 TV
      6               TV                 IN                 ZA
      7               UK                 TV                  IN
      8               FR                 DE                 UK
      9               PE                 PE                 DE
     10              GOV                 FR                 GOV
     11              EDU                 ZA                 EDU
     12               DE                GOV                 RU
     13               RU                 NU                 FR
     14               US                EDU                 US
     15               NU                 RU                 PE
     16               IT                 IT                 CN
     17               AR                 US                  IT
     18              MX                  MX                 CA
     19               CA                 CN                 SG
     20              INFO                AR                 MX
     21               CO                 CA                 INFO
     22               BR                 IE                 NU
     23               CN                INFO                AR
     24               ES                 TH                 PL
     25               FM                 ES                 FM
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009
Stateoftheweb q4-2009

More Related Content

What's hot

Sap r3 basic_training_finance_en_v5
Sap r3 basic_training_finance_en_v5Sap r3 basic_training_finance_en_v5
Sap r3 basic_training_finance_en_v5Casst346
 
Parallels Plesk Panel 9 Client's Guide
Parallels Plesk Panel 9 Client's GuideParallels Plesk Panel 9 Client's Guide
Parallels Plesk Panel 9 Client's Guidewebhostingguy
 
Ecdl v5 module 4 print
Ecdl v5 module 4 printEcdl v5 module 4 print
Ecdl v5 module 4 printMichael Lew
 
Ecdl v5 module 3 print
Ecdl v5 module 3 printEcdl v5 module 3 print
Ecdl v5 module 3 printMichael Lew
 
Functional Specs Short
Functional Specs  ShortFunctional Specs  Short
Functional Specs Shortlisalugo
 
Parallels Plesk Panel 9 Reseller's Guide
Parallels Plesk Panel 9 Reseller's GuideParallels Plesk Panel 9 Reseller's Guide
Parallels Plesk Panel 9 Reseller's Guidewebhostingguy
 
Manual smart notebook se mac
Manual smart notebook se macManual smart notebook se mac
Manual smart notebook se macecoiote
 
Plesk Sitebuilder 4.5 for Linux/Unix Wizard User's Guide
Plesk Sitebuilder 4.5 for Linux/Unix Wizard User's GuidePlesk Sitebuilder 4.5 for Linux/Unix Wizard User's Guide
Plesk Sitebuilder 4.5 for Linux/Unix Wizard User's Guidewebhostingguy
 
Verio Web Hosting Virtual Server Handbook
Verio Web Hosting Virtual Server HandbookVerio Web Hosting Virtual Server Handbook
Verio Web Hosting Virtual Server Handbookwebhostingguy
 
8 2-sp1 administering-broker
8 2-sp1 administering-broker8 2-sp1 administering-broker
8 2-sp1 administering-brokerNugroho Hermanto
 
CPanel 1.01 User Guide
CPanel 1.01 User GuideCPanel 1.01 User Guide
CPanel 1.01 User Guidewebhostingguy
 
Linux for professional
Linux for professionalLinux for professional
Linux for professionalBennethObilor
 
hp StorageWorks host bus adapter for Windows and Linux ...
hp StorageWorks host bus adapter for Windows and Linux ...hp StorageWorks host bus adapter for Windows and Linux ...
hp StorageWorks host bus adapter for Windows and Linux ...webhostingguy
 
Zeta Producer 9 CMS online manual
Zeta Producer 9 CMS online manualZeta Producer 9 CMS online manual
Zeta Producer 9 CMS online manualUwe Keim
 

What's hot (20)

Sap r3 basic_training_finance_en_v5
Sap r3 basic_training_finance_en_v5Sap r3 basic_training_finance_en_v5
Sap r3 basic_training_finance_en_v5
 
Parallels Plesk Panel 9 Client's Guide
Parallels Plesk Panel 9 Client's GuideParallels Plesk Panel 9 Client's Guide
Parallels Plesk Panel 9 Client's Guide
 
Ecdl v5 module 4 print
Ecdl v5 module 4 printEcdl v5 module 4 print
Ecdl v5 module 4 print
 
R Ints
R IntsR Ints
R Ints
 
Ecdl v5 module 3 print
Ecdl v5 module 3 printEcdl v5 module 3 print
Ecdl v5 module 3 print
 
Functional Specs Short
Functional Specs  ShortFunctional Specs  Short
Functional Specs Short
 
PlayBook userguide
PlayBook userguidePlayBook userguide
PlayBook userguide
 
Parallels Plesk Panel 9 Reseller's Guide
Parallels Plesk Panel 9 Reseller's GuideParallels Plesk Panel 9 Reseller's Guide
Parallels Plesk Panel 9 Reseller's Guide
 
End note
End noteEnd note
End note
 
Sap In-Memory IBM
Sap In-Memory IBMSap In-Memory IBM
Sap In-Memory IBM
 
Manual smart notebook se mac
Manual smart notebook se macManual smart notebook se mac
Manual smart notebook se mac
 
MS Word 2000
MS Word 2000MS Word 2000
MS Word 2000
 
Plesk Sitebuilder 4.5 for Linux/Unix Wizard User's Guide
Plesk Sitebuilder 4.5 for Linux/Unix Wizard User's GuidePlesk Sitebuilder 4.5 for Linux/Unix Wizard User's Guide
Plesk Sitebuilder 4.5 for Linux/Unix Wizard User's Guide
 
Verio Web Hosting Virtual Server Handbook
Verio Web Hosting Virtual Server HandbookVerio Web Hosting Virtual Server Handbook
Verio Web Hosting Virtual Server Handbook
 
8 2-sp1 administering-broker
8 2-sp1 administering-broker8 2-sp1 administering-broker
8 2-sp1 administering-broker
 
Google Search Quality Rating Program General Guidelines 2011
Google Search Quality Rating Program General Guidelines 2011Google Search Quality Rating Program General Guidelines 2011
Google Search Quality Rating Program General Guidelines 2011
 
CPanel 1.01 User Guide
CPanel 1.01 User GuideCPanel 1.01 User Guide
CPanel 1.01 User Guide
 
Linux for professional
Linux for professionalLinux for professional
Linux for professional
 
hp StorageWorks host bus adapter for Windows and Linux ...
hp StorageWorks host bus adapter for Windows and Linux ...hp StorageWorks host bus adapter for Windows and Linux ...
hp StorageWorks host bus adapter for Windows and Linux ...
 
Zeta Producer 9 CMS online manual
Zeta Producer 9 CMS online manualZeta Producer 9 CMS online manual
Zeta Producer 9 CMS online manual
 

Viewers also liked

Präsentation über die Entwicklungshilfeorganisation People help People - One ...
Präsentation über die Entwicklungshilfeorganisation People help People - One ...Präsentation über die Entwicklungshilfeorganisation People help People - One ...
Präsentation über die Entwicklungshilfeorganisation People help People - One ...PHPOW
 
Company presentation People help People - One World
Company presentation People help People - One WorldCompany presentation People help People - One World
Company presentation People help People - One WorldPHPOW
 
Concept People help People - One World
Concept  People help People - One WorldConcept  People help People - One World
Concept People help People - One WorldPHPOW
 
#dmu14 Remarketing: Cómo optimizar tus tasas de conversión
#dmu14 Remarketing: Cómo optimizar tus tasas de conversión#dmu14 Remarketing: Cómo optimizar tus tasas de conversión
#dmu14 Remarketing: Cómo optimizar tus tasas de conversiónIgni
 
Misra, D.C.(2008) IFCs&Egov_ IIPA_ 28.2.08
Misra, D.C.(2008) IFCs&Egov_ IIPA_ 28.2.08Misra, D.C.(2008) IFCs&Egov_ IIPA_ 28.2.08
Misra, D.C.(2008) IFCs&Egov_ IIPA_ 28.2.08Dr D.C. Misra
 
People help People - One World / Aktueller Newsletter 3/2010
People help People - One World / Aktueller Newsletter 3/2010People help People - One World / Aktueller Newsletter 3/2010
People help People - One World / Aktueller Newsletter 3/2010PHPOW
 
The Outcome Economy
The Outcome EconomyThe Outcome Economy
The Outcome EconomyHelge Tennø
 

Viewers also liked (8)

Präsentation über die Entwicklungshilfeorganisation People help People - One ...
Präsentation über die Entwicklungshilfeorganisation People help People - One ...Präsentation über die Entwicklungshilfeorganisation People help People - One ...
Präsentation über die Entwicklungshilfeorganisation People help People - One ...
 
Company presentation People help People - One World
Company presentation People help People - One WorldCompany presentation People help People - One World
Company presentation People help People - One World
 
Xd 8550-6000
Xd 8550-6000Xd 8550-6000
Xd 8550-6000
 
Concept People help People - One World
Concept  People help People - One WorldConcept  People help People - One World
Concept People help People - One World
 
#dmu14 Remarketing: Cómo optimizar tus tasas de conversión
#dmu14 Remarketing: Cómo optimizar tus tasas de conversión#dmu14 Remarketing: Cómo optimizar tus tasas de conversión
#dmu14 Remarketing: Cómo optimizar tus tasas de conversión
 
Misra, D.C.(2008) IFCs&Egov_ IIPA_ 28.2.08
Misra, D.C.(2008) IFCs&Egov_ IIPA_ 28.2.08Misra, D.C.(2008) IFCs&Egov_ IIPA_ 28.2.08
Misra, D.C.(2008) IFCs&Egov_ IIPA_ 28.2.08
 
People help People - One World / Aktueller Newsletter 3/2010
People help People - One World / Aktueller Newsletter 3/2010People help People - One World / Aktueller Newsletter 3/2010
People help People - One World / Aktueller Newsletter 3/2010
 
The Outcome Economy
The Outcome EconomyThe Outcome Economy
The Outcome Economy
 

Similar to Stateoftheweb q4-2009

Similar to Stateoftheweb q4-2009 (20)

Ctfile
CtfileCtfile
Ctfile
 
C01508406
C01508406C01508406
C01508406
 
Amdin iws7 817-2179-10
Amdin iws7 817-2179-10Amdin iws7 817-2179-10
Amdin iws7 817-2179-10
 
R data
R dataR data
R data
 
M Daemon E Mail Server Manual
M Daemon E Mail Server ManualM Daemon E Mail Server Manual
M Daemon E Mail Server Manual
 
cynapspro endpoint data protection - user guide
cynapspro endpoint data protection -  user guidecynapspro endpoint data protection -  user guide
cynapspro endpoint data protection - user guide
 
Gdfs sg246374
Gdfs sg246374Gdfs sg246374
Gdfs sg246374
 
Xi iii plus_userguide
Xi iii plus_userguideXi iii plus_userguide
Xi iii plus_userguide
 
R Data
R DataR Data
R Data
 
Java script tools guide cs6
Java script tools guide cs6Java script tools guide cs6
Java script tools guide cs6
 
B035-2447-220K.pdf
B035-2447-220K.pdfB035-2447-220K.pdf
B035-2447-220K.pdf
 
60950106 basis-latest-till-interview-questions
60950106 basis-latest-till-interview-questions60950106 basis-latest-till-interview-questions
60950106 basis-latest-till-interview-questions
 
SEAMLESS MPLS
SEAMLESS MPLSSEAMLESS MPLS
SEAMLESS MPLS
 
MongoDB Use Case - Mobile App Backend
MongoDB Use Case - Mobile App BackendMongoDB Use Case - Mobile App Backend
MongoDB Use Case - Mobile App Backend
 
Z4 mz6musersguide
Z4 mz6musersguideZ4 mz6musersguide
Z4 mz6musersguide
 
C sharp programming
C sharp programmingC sharp programming
C sharp programming
 
C sharp programming[1]
C sharp programming[1]C sharp programming[1]
C sharp programming[1]
 
Tx16 wx user manual
Tx16 wx user manualTx16 wx user manual
Tx16 wx user manual
 
Novell login documentation and troubleshooting
Novell login documentation and troubleshootingNovell login documentation and troubleshooting
Novell login documentation and troubleshooting
 
Fortigate utm-40-mr1
Fortigate utm-40-mr1Fortigate utm-40-mr1
Fortigate utm-40-mr1
 

Stateoftheweb q4-2009

  • 1. State  of  the  Web  -­‐  Q4  2009 A  View  of  the  Web  From  an  End  User’s  Perspec:ve Zscaler  Labs Abstract Attackers  are  no  longer  targeting  web  and  email  servers.  Today,  they  are  attacking   enterprises  from  the  inside  out,  by  <irst  compromising  end  user  systems  and  then   leveraging  them  to  gain  access  to  con<idential  data.  As  such  it  is  imperative  that   organizations  have  an  understanding  of  what  is  happening  on  the  web.  As  a   Security-­‐as-­‐a-­‐Service  vendor,  Zscaler  has  a  unique  perspective  on  web  traf<ic.  With   millions  of  end  users  traversing  the  web  through  Zscaler’s  global  network  of  web   gateways,  we  are  able  to  better  understand  both  how  users  are  interacting  with   web  based  resources  and  how  attackers  may  be  targeting  end  users.  In  this,  our   <irst  quarterly  ‘State  of  the  Web’  report,  we  provide  a  window  into  the  web  from  an   end  user’s  perspective.
  • 2. Table  of  Contents Overview  .....................................................................................................................................3 Web  Traf/ic  Statistics  ..............................................................................................................3 Web  Server  Statistics  .......................................................................................................................3 TLDs  by  Unique  Domain  Visited  ...............................................................................................................4 TLDs  by  Total  Transactions   .........................................................................................................................6 Transaction  to  Domain  Ratio  .....................................................................................................................6 Top  Domains  Visited  ......................................................................................................................................8 CIDR  Block  Distribution  ...............................................................................................................................9 ASN  Distribution  ...........................................................................................................................................11 Geography   ........................................................................................................................................................12 File  Types  .........................................................................................................................................................15 Request  Method  ............................................................................................................................................15 Response  Code  ...............................................................................................................................................16 Web  Browser  Statistics   ..................................................................................................................17 Browser  Version   ............................................................................................................................................17 User  Statistics  ..................................................................................................................................19 URL  Categorization  ......................................................................................................................................19 Search  Engines  ...............................................................................................................................................19 Social  networking  .........................................................................................................................................20 File  Sharing  ......................................................................................................................................................20 Government  ....................................................................................................................................................21 Retail  ..................................................................................................................................................................21 Security  Statistics  ..................................................................................................................22 Threats  ...............................................................................................................................................22 Malware  By  IP  Address  ..............................................................................................................................22 Malware  by  Country  ....................................................................................................................................22 Phishing  ............................................................................................................................................................23 Malicious  Domains  .......................................................................................................................................24 Anonymizers  ...................................................................................................................................................25 Botnets  ..............................................................................................................................................................25 Traf/ic  ..................................................................................................................................................26 Bogon  IP  space  ...............................................................................................................................................26 Conclusion  ................................................................................................................................28 Appendix  ..................................................................................................................................29 TLD  Breakdown  ..............................................................................................................................29 Monthly  Summary  –  Top  TLDs  Visited  ................................................................................................29 Monthly  Summary  -­‐  Unique  Domains  Per  TLD  ................................................................................30 Categorization  Breakdown  ..........................................................................................................31 .COM  Breakdown  by  Category  ................................................................................................................31 .NET  Breakdown  by  Category  .................................................................................................................34 .ORG  Breakdown  by  Category  .................................................................................................................37 .INFO  Breakdown  by  Category  ................................................................................................................40 Top  Search  Queries  .........................................................................................................................49
  • 3. Overview Our  goal  in  producing  this  report  is  to  better  understand  traf<ic  on  the  web  today.   Security  and  IT  teams  across  organizations  are  tasked  with  managing  the  traf<ic  of   end  users  on  their  networks.  That  can  involve  restricting  access  for  various  business   purposes  and  protecting  end  users  from  external  threats.  This  is  a  tall  order  given   the  increasing  ease  of  access  to  web  based  services  that  often  permit  users  to  bypass   traditional  controls  -­‐  whether  accessing  corporate  resources  from  personal  devices   such  as  smart  phones  or  setting  up  applications  by  leveraging  cloud  based   resources. Zscaler  is  in  a  unique  position  to  observe  trends  in  web  traf<ic.  As  a  Security-­‐as-­‐a-­‐ Services  vendor,  Zscaler’s  network  of  web  gateways  continually  inspects  traf<ic  for   millions  of  end  users  around  the  globe.   There  are  a  number  of  great  reports  available  today  from  a  variety  of  organizations   to  help  us  better  understand  web  traf<ic.  However,  the  majority  of  such  reports  tend   to  focus  on  the  server  side  of  the  equation.  They  tend  to  look  at  the  technology  that   has  been  deployed  to  deliver  web  content  and  associated  security  issues  in  web   applications.  We  feel  that  there  is  a  need  to  better  understand  the  client  side  of  the   equation  -­‐  what  are  end  users  doing  on  the  web  and  how  are  attackers  targeting   them?  The  latter  part  of  this  question  is  especially  important  as  attackers  have   clearly  shifted  away  from  attacking  web  and  email  servers  to  targeting  end  users.   They  understand  that  end  user  systems  tend  to  represent  the  weakest  link  in  the   security  chain  and  they  are  exploiting  that  weakness  with  increasing  ef<iciency.  We   can  better  defend  against  such  attacks  by  better  understanding  exactly  what  is   occurring  on  the  web  and  it  is  our  hope  that  this  report  will  help  to  shed  some  light   on  that  very  topic. Web  Traffic  Sta0s0cs Web  Server  Sta0s0cs Zscaler  customers  visited  several  million  web  servers  during  the  4th  quarter  of   2009.  One  interesting  technique  for  visualizing  the  IP  addresses  of  the  web  servers   visited  is  through  a  heatmap.  The  below  graphic  was  generated  from  the   Measurement  Factory  software1  and  “uses  a  12th-­‐order  Hilbert  curve2  to  represent   the  entire  IPv4  address  space”.  In  the  graphic  below,  IP  addresses  visited  are   represented  by  white  pixels,  while  addresses  not  visited  are  displayed  as  black   pixels.  Non-­‐routable  or  reserved  space  is  identi<ied  in  gray  and  where  appropriate,   we  have  indicated  what  that  space  is  used  for.  It’s  a  fascinating  view  which  exposes   just  how  vast  the  Internet  truly  is.  Even  when  analyzing  traf<ic  from  millions  of  users   1 http://maps.measurement-factory.com/software/ipv4-heatmap.1.html 2A Hilbert Curve is a space filling curve that visits every point in a grid (in our case a 2^12 x 2^12 grid).
  • 4. over  the  course  of  three  months,  it  can  be  seen  that  much  of  the  Internet  remains   untouched. Hilbert  Curve  -­‐  All  Q4  2009  traffic  by  IP  address TLDs  by  Unique  Domain  Visited .com,  .org,  and  .net  top-­‐level  domains  (TLDs)   consistently  made  up  the  bulk  of  the  unique   Other 10% domains  visited  each  month.  .com  traf<ic  made  up   org 80.11%  of  the  unique  domains  visited  during  the   4% net quarter.  .net  had  over  4.96%  and  .org  accounted  for   5% 4.45%.  The  chart  below  shows  the  next  10  largest   TLDs  that  make  up  about  80%  of  the  remaining   11%  of  the  unique  domains  visited  in  Q4  of  2009. com 80%
  • 5. Top 10 TLDs By Unique Domain Per Month (Excluding .com/.net/.org) 0% 0.50% 1.00% 1.50% 2.00% 1.25% ru 1.1% 1.50% 1.04% uk 1.5% 1.22% 1.03% au 1.3% 1.30% 1.23% edu 1.3% 1.11% 0.74% de 1.0% 1.07% 0.59% info 0.6% 0.63% 0.63% us 0.6% 0.57% 0.56% fr 0.6% 0.49% 0.63% in 0.5% 0.47% 0.39% ca 0.5% 0.44% October November December There  were  a  number  of  similarities  with  a  few  <luctuations  within  the  top  10  TLDs   with  unique  domains  visited  from  month-­‐to-­‐month: • .ru  was  the  4th  most  popular  TLD  by  unique  domain  visited  in  October  and   December,  however  it  dropped  to  the  7th  spot  in  November. • .au,  .uk,  and  .edu  make  up  the  5-­‐7  spots,  with  the  exception  of  November   when  .uk  beat  out  .ru  for  the  4th  spot. The  chart  below  shows  the  breakdown  of  TLDs  based  on  total  number  of   transactions  as  opposed  to  unique  domains.  This  view  would  favor  those  TLDs   hosting  popular  sites  which  receive  higher  volumes  of  overall  traf<ic.
  • 6. TLDs  by  Total  Transac0ons Top 10 TLDs By Transaction Per Month (Excluding .com/.net) 0% 0.50% 1.00% 1.50% 2.00% 1.57% org 1.54% 1.44% 0.66% au 0.97% 0.78% 0.62% in 0.38% 0.41% 0.36% tv 0.35% 0.55% 0.29% uk 0.44% 0.41% 0.16% de 0.23% 0.26% 0.22% gov 0.19% 0.25% 0.24% fr 0.21% 0.18% 0.21% edu 0.18% 0.20% 0.23% pe 0.22% 0.16% October November December Transac0on  to  Domain  Ra0o The  data  from  the  top  TLDs  by  unique  domain  and  top  TLDs  by  transaction  can  be   combined  to  <ind:   • The  TLDs  with  the  highest  ratio  of  transactions  to  domains  –  indicating  a   large  number  of  transactions  across  a  small  subset  of  domains.    In  other   words,  there  are  only  a  few  unique  domains  in  the  TLD  that  make  it  popular. • The  TLDs  with  the  lowest  ratio  of  transactions  to  domains  –  indicating  a   number  of  domains  among  which  the  transactions  are  spread  out.  In  other   words,  the  unique  domains  that  have  a  small  number  of  visits  or  transactions.  
  • 7. October November December Rank TLD Ratio TLD Ratio TLD Ratio 1 nu 5063 nu 8083 net 3737 2 net 3617 net 3428 nu 2824 3 ly 2140 ly 1792 ly 1699 4 tv 1719 tv 1568 tv 1692 5 pe 1326 id 1267 fm 1307 6 fm 1140 pe 1159 lan 1260 7 in 803 lan 1153 pe 1228 8 com 726 fm 968 com 799 9 it 707 ir 807 in 765 10 id 677 com 702 im 713 11 aero 676 it 678 pf 701 12 hn 662 in 655 it 633 13 tr 655 th 622 gov 592 14 au 520 au 584 th 587 15 su 500 tr 531 vn 546 16 im 483 gr 515 au 517 17 gov 480 dk 503 ir 485 18 ke 422 local 471 za 463 19 ph 409 hn 398 hn 461 20 ec 400 gov 392 tr 436 21 int 391 mx 389 mx 417 22 co 355 ke 374 sg 382 23 fr 341 io 368 ke 380 24 th 330 co 331 va 372 25 mx 315 ec 325 ec 368 Well  utilized,  generic  TLDs  (gTLD),  such  as  .com,  will  have  a  high  ratio  because   domains  like  Google,  Facebook,  Amazon,  Yahoo,  Microsoft,  MySpace,  Twitter,  etc.   contain  a  large  number  of  the  transactions  to  that  TLD.  This  is  however  offset  to  a   certain  extent  because  there  are  also  a  large  number  of  popular  domains  on  these   gTLDs  and  these  unique  domains  will  lower  the  ratio  somewhat,  though  it  remains   relatively  high  overall.  For  example,  October  –  December  2009  saw  .com  ratios  of   726:1,  702:1,  and  799:1  respectively. It  is  interesting  to  further  analyze  domain  results  for  less  popular  TLDs  and  those   that  had  a  higher  ratio  than  the  gTLDs,  both  from  a  statistical  and  trending   perspective  as  well  as  from  a  security  perspective.  Miscreants  frequently  register   domains  with  TLDs  that  are  less  in  demand  because  they  are  cheaper,  and  in  some   cases  the  particular  domain  registry  (maintainer  of  the  TLD)  and/or  registrar   (maintainer  of  the  domain  record)  will  have  poor  abuse  handling  procedures.   Additionally,  the  registry  and/or  registrar  may  either  be  complicit  in  the  illegal   activity  or  be  in  a  jurisdiction/country  with  a  legal  system  that  protects  the  domain   from  being  de-­‐registered  or  having  the  registration  information  shared  with  law   enforcement.  TLDs  with  a  high-­‐ratio  of  transactions  per  unique  domain  per  TLD  
  • 8. have  one  or  more  domains  with  a  large  number  of  transactions.  It  is  interesting  to   sift  through  the  records  to  explain  the  high-­‐ratio  TLDs.  They  may  be  the  result  of  a   malicious  command  and  control  (C&C)  or  information  drop  server  that  has  a  large   number  of  transactions  beaconing  to  the  domain’s  server,  or  it  could  be  something   benign,  such  as  a  popular  social  networking  site  in  a  particular  country. One  such  example  of  a  benign  domain  within  a  TLD  that  bubbled  to  the  top  was  .ly.     This  domain  had  a  ratio  of  2140:1,  1792:1,  and  1699:1  in  the  October  –  December   timeframe.  These  ratios  were  more  than  double  the  ratios  that  .com  had  during   these  months.  This  high  ratio  is  explained  by  this  TLD  being  relatively  unpopular  as   far  as  unique  domains  go,  but  having  a  large  number  of  transactions  to  a  popular   domain  -­‐  namely  bit.ly,  a  popular  URL  shortening  service. The  .nu  TLD  had  even  higher  ratios  of  5063:1,  8083:1,  and  2824:1  in  Q4  2009.   The  .nu  TLD  is  assigned  to  the  island  state  of  Niue,  and  Wikipedia  states  that  the  TLD   “is  particularly  popular  in  Sweden,  Denmark,  the  Netherlands  and  Belgium,  as  nu  is   the  word  for  ‘now’  in  Swedish,  Danish,  and  Dutch.”  While  the  domain  may  be  popular   for  these  countries,  our  ratio  shows  that  a  relatively  small  number  of  domains  are   dominating  the  transactions  for  this  TLD. Running  a  query  against  the  Zscaler  NanoLogs  for  the  .nu  domains  and  count  of   transactions,  yielded  a  large  percentage  of  the  transactions  to  the  domain:   cvnxus.mine.nu.  The  transactions  to  the  domain  appear  as:   hxxp://  cvnxus.mine.nu:53/30080000 Further  analysis  revealed  that  there  were  several  bot  infected  hosts  that  were   beaconing  TCP  ACK  packets  to  this  host.    Zscaler  has  since  noti<ied  and  assisted   impacted  customers.  A  separate  white  paper  detailing  this  analysis  will  be  released. Top  Domains  Visited Many  of  the  most  visited  domains  are  actually  those  that  operate  behind  the  scenes.   liveperson.net  for  example  is  a  real-­‐time  support  tool  used  by  a  variety  of  large   online  retail  and  services  companies  such  as  Bank  of  America,  AT&T  and  IBM3.  As   such,  when  receiving  email  and  chat  based  customer  support  at  such  companies   certain  traf<ic  is  actually  redirected  to  the  liverperson.net  domain.  Top  domains  are   calculated  based  on  the  total  number  of  transactions.  As  such,  sites  delivering   images,  streaming  content  or  requiring  frequent  communication  of  some  form  tend   to  score  higher.  Advertising  based  traf<ic  was  very  prevalent  with  ad  management   platforms  such  as  doubleclick.net  and  yieldmanager.com,  both  landing  in  the  top  10.   Google,  Yahoo!  and  Facebook  all  ranked  high,  as  did  domains  owned  and  managed   by  them.  <bcdn.net  and  yimg.com  serve  up  Facebook  and  Yahoo!  content   respectively.  google-­‐analytics.com,  a  Google  tool  for  tracking  site  visitors  receives   signi<icant  traf<ic  due  to  the  fact  that  links  to  the  domain  are  posted  on  numerous   third  party  sites. 3 http://solutions.liveperson.com/company/customers/
  • 9. Top 10 Domains Visited By Month Q4 2009 0% 7.50% 15.00% 22.50% 30.00% liveperson.net google.com doubleclick.net fbcdn.net yahoo.com yimg.com facebook.com google-analytics.com yieldmanager.com login.icq.com October November December CIDR  Block  Distribu0on CIDR  notation  is  a  way  of  writing  a  block  of  IP  addresses,  where  the  suf<ix  number  is   the  number  of  bits  to  include  from  the  IP  for  the  block4 .  For  example: • 192.168.1.0/24  is  the  IP  block:  192.168.1.0-­‐192.168.1.255 • 192.168.0.0/16  is  the  IP  block:  192.168.0.0-­‐192.168.255.255 • 192.0.0.0/8  is  the  IP  block:  192.0.0.0-­‐192.255.255.255 The  chart  below  shows  the  top  25  most  popular,  highly  utilized  IP  blocks  based  on   Zscaler  customer  traf<ic.  These  results  are  displayed  in  three  ways:  (1)  a  narrow,  /24   IP  block,  viewpoint,  (2)  a  middle,  /16  IP  block,  viewpoint,  and  (3)  a  broader,  /8  IP   block,  viewpoint. The  narrow,  /24  IP  block,  viewpoint  is  largely  comprised  of  popular  end-­‐user  sites/ services  that  are  distributed  across  their  IP  block.  The  4th  quarter  included  some  of   4 http://en.wikipedia.org/wiki/CIDR_notation
  • 10. the  busiest  shopping  months  of  the  year.  This,  combined  with  Amazon's  utilization   of  their  IP  blocks  (e.g.,  their  EC2  service),  accounted  for  Amazon  having  the  top  10  / 24  IP  blocks  by  number  of  unique  IPs  visited.  MySpace  and  Vkontakte  are  social   networking  sites  that  seem  to  distribute  their  user  load  and/or  content  among  a   number  of  web  server  IPs  in  their  block. The  middle  /16  IP  block,  displays  some  of  the  more  popular  hosting  and  service   providers  by  unique  IPs  visited,  such  as,  1&1,  Digital  United,  Taiwan  Fixed  Network,   and  HiNet.    It  is  interesting  that  when  looking  at  the  most  popular  IP  blocks  from  a   middle  aggregation  point,  /16  IP  blocks,  more  Asia  based  IP  blocks  bubble  to  the   top.  From  smaller  (/24  IP  blocks)  and  larger  (/8  IP  blocks)  IP  aggregation  points,   more  United  States  based,  ARIN  space  <inds  its  way  into  the  top  25  blocks  by  unique   IP  visited.  This  suggests  that  Asian  /  APNIC  service  and  hosting  providers  may   largely  be  constructed  of  /16  or  similar  sized  blocks.   /24 CIDR Block /16 CIDR Block /8 CIDR Block Rank Range Organization Range Organization Range 1 216.137.37.0/24 Amazon 74.208.0.0/16 1&1 Internet Inc. 74.208.0.0/8 2 216.137.39.0/24 Amazon 123.204.0.0/16 Digital United 69.0.0.0/8 3 216.137.41.0/24 Amazon 124.8.0.0/16 Taiwan Fixed Network 216.0.0.0/8 4 216.137.45.0/24 Amazon 114.44.0.0/16 HiNet 66.0.0.0/8 5 216.137.47.0/24 Amazon 219.85.0.0/16 Sony Network Taiwan 74.0.0.0/8 6 216.137.55.0/24 Amazon 124.218.0.0/16 Asia Pacific On-line 208.0.0.0/8 7 216.137.59.0/24 Amazon 122.121.0.0/16 HiNet 64.0.0.0/8 8 216.137.53.0/24 Amazon 220.136.0.0/16 HiNet 72.0.0.0/8 9 216.137.43.0/24 Amazon 125.230.0.0/16 HiNet 67.0.0.0/8 10 216.137.61.0/24 Amazon 114.47.0.0/16 HiNet 61.0.0.0/8 11 63.135.88.0/24 MySpace 112.104.0.0/16 Digital United 218.0.0.0/8 12 216.137.35.0/24 Amazon 59.117.0.0/16 HiNet 209.0.0.0/8 13 91.192.55.0/24 spamfighter.com 118.160.0.0/16 HiNet 118.0.0.0/8 14 93.186.229.0/24 Vkontakte.ru 74.125.0.0/16 Google Inc. 174.0.0.0/8 15 70.35.16.0/24 Netfirms, Inc. 218.172.0.0/16 HiNet 65.0.0.0/8 16 93.186.230.0/24 Vkontakte.ru 69.192.0.0/16 Akamai Technologies 122.0.0.0/8 17 64.71.33.0/24 affinity.com 96.17.0.0/16 Akamai Technologies 207.0.0.0/8 18 69.89.31.0/24 bluehost.com 96.6.0.0/16 Akamai Technologies 87.0.0.0/8 19 65.54.81.0/24 Microsoft.com 118.171.0.0/16 HiNet 220.0.0.0/8 20 64.12.24.0/24 aol.net 219.81.0.0/16 Taiwan Fixed Network 124.0.0.0/8 21 124.218.196.0/24 Asia Pacific On-line 114.40.0.0/16 HiNet 125.0.0.0/8 22 124.218.194.0/24 Asia Pacific On-line 219.84.0.0/16 Sony Network Taiwan 219.0.0.0/8 23 124.218.198.0/24 Asia Pacific On-line 218.163.0.0/16 HiNet 59.0.0.0/8 24 124.218.200.0/24 Asia Pacific On-line 61.31.0.0/16 Taiwan Fixed Network 96.0.0.0/8 25 124.218.202.0/24 Asia Pacific On-line 114.43.0.0/16 HiNet 82.0.0.0/8 To  get  a  clearer  picture  of  actual  organizations  with  a  large  number  of  visited  web   servers  (unique  web  server  IPs),  a  chart  was  created  breaking  out  unique  IPs  visited   per  autonomous  system.  An  autonomous  system  (AS)  is  a  collection  of  connected  IP  
  • 11. blocks  under  the  control  a  group/organization.    The  <irst  and  third  most  popular  ASs   are  Asian,  which  correlates  with  our  previous  statement. ASN  Distribu0on Rank ASN Organization Percentage 1 AS3462 HINET Data Communication Business Group 13.36% 2 AS21844 ThePlanet.com Internet Services, Inc. 2.31% 3 AS9924 Taiwan Fixed Network, Telco and Network Service Provider. 1.53% 4 AS2914 NTT America, Inc. 1.36% 5 AS8560 1&1 Internet AG 1.33% 6 AS7132 AT&T Internet Services 1.27% 7 AS4780 Digital United Inc. 1.21% 8 AS33070 Rackspace.com, Ltd. 1.02% 9 AS4134 CHINANET-BACKBONE No.31,Jin-rong Street 1.01% 10 AS18182 Sony Network Taiwan Limited 1.01% 11 AS36351 SoftLayer Technologies Inc. 0.97% 12 AS26347 New Dream Network, LLC 0.95% 13 AS26496 GoDaddy.com, Inc. 0.95% 14 AS3269 TELECOM ITALIA 0.86% 15 AS209 Qwest Communications Company, LLC 0.82% 16 AS3356 Level 3 Communications 0.63% 17 AS20940 Akamai Technologies European AS 0.62% 18 AS15169 Google Inc. 0.53% 19 AS16276 OVH 0.50% 20 AS32244 Liquid Web, Inc. 0.48% 21 AS3215 France Telecom - Orange 0.43% 22 AS12322 PROXAD AS for Proxad/Free ISP 0.41% 23 AS3549 Global Crossing Ltd. 0.33% 24 AS22822 Limelight Networks, Inc. 0.33% 25 AS7482 Asia Pacific On-line Service Inc. 0.27%
  • 12. Geography The  majority  of  requested  web  content  resides  on  servers  located  in  the  United   States.  With  the  exception  of  a  spike  in  October  and  part  of  November  for  content   located  in  Taiwan,  traf<ic  not  destined  for  non-­‐US  based  web  sites  was  fairly  evenly   distributed  across  servers  located  primarily  in  a  variety  of  countries  in  Europe  and   Asia. Top 10 Destinations By Country Q4 2009 0% 15.00% 30.00% 45.00% 60.00% United States Taiwan Germany France United Kingdom China Canada Italy Russian Federation Japan October November December Country October November December United States 35.73% 44.42% 55.74% Taiwan 34.22% 10.41% 3.15% Germany 2.76% 4.36% 4.32% France 1.82% 5.27% 2.84% United Kingdom 2.21% 3.42% 3.83% China 2.01% 2.70% 2.66% Canada 1.97% 2.54% 2.71% Italy 2.55% 1.63% 0.87% Russian Federation 1.42% 2.09% 1.83% Japan 1.64% 1.65% 1.61%
  • 13. Top 10 Destinations By Region Q4 2009 0% 10.00% 20.00% 30.00% 40.00% Taipei California Texas Massachusetts Pennsylvania Arizona New York Illinois Taiwan Florida October November December Region October November December Taipei 30.14% 9.10% 2.67% California 6.94% 7.23% 8.87% Texas 4.87% 6.40% 8.33% Massachusetts 1.98% 2.71% 3.61% Pennsylvania 1.55% 2.06% 2.62% Arizona 1.53% 2.00% 2.65% New York 1.41% 1.92% 2.17% Illinois 1.30% 1.71% 2.15% Taiwan 3.26% 1.04% Florida 1.25% 1.64% 2.08% From  a  regional  perspective,  Taipei  hosted  much  of  the  Taiwanese  based  content   which  accounted  for  the  surge  in  October  and  November.  As  for  US  traf<ic,  both   California  and  Texas  account  for  the  bulk  of  content,  with  the  remainder  tending  to   be  located  on  the  East  Coast.
  • 14. Top 10 Destinations By City Q4 2009 0% 10.00% 20.00% 30.00% 40.00% Taipei Houston Cambridge San Antonio Dallas Scottsdale Englewood Seattle Moscow Brea October November December City October November December Taipei 30.14% 9.10% 2.67% Houston 1.87% 2.48% 3.21% Cambridge 1.41% 1.98% 2.79% San Antonio 1.08% 1.40% 1.85% Dallas 0.98% 1.30% 1.73% Scottsdale 0.75% 1.00% 1.34% Englewood 0.77% 0.97% 1.30% Seattle 0.71% 0.91% 1.17% Moscow 0.71% 0.99% 0.93% Brea 0.66% 0.83% 1.12%
  • 15. File  Types Many  assume  that  web  traf<ic  is   Top 10 File Types Q4 2009 dominated  by  HTML  content.  While   that  may  have  been  true  a  decade   ago,  the  media  rich,  dynamic  web   0% 7.50% 15.00% 22.50% applications  available  today  are   30.00% <illed  with  images,  formatting   jpeg 28.74% elements,  data  and  active  content.   gif 28.68% For  the  4th  quarter  JPEG  (28.74%)   gz 24.40% and  GIF  (28.68%)  images  alone   png 6.25% accounted  for  more  than  half  of  the   js 3.95% total  number  of  transactions.  This  is   css 1.82% a  testament  to  the  visual  nature  of   swf 1.14% the  web.  JavaScript,  the  ‘work  horse’   xml 0.72% txt 0.66% of  modern,  user-­‐friendly  web   jpg 0.57% applications  was  responsible  for   only  3.95%  of  transactions.  HTML   <iles  fell  just  outside  of  the  top  10   and  drove  0.57%  of  traf<ic. Request  Method Predictably,  GET  requests  account   INVALID 0.04% 0.24% for  the  majority  of  traf<ic.  Generally   0% speaking,  GET  and  POST  requests   86.58% are  the  most  often-­‐used   GET 83.46% 96.29% communication  methods  employed   7.18% by  web  applications,  the  difference   POST 15.16% being  that  a  GET  request  passes   3.67% 0.14% request  variables  within  the  URL   HEAD 0.07% itself  while  POST  requests  pass   0% variables  as  a  portion  of  the  request   0% MOVE 0% header.  Each  approach  has   0% advantages  and  disadvantages  but   6.04% CONNECT 1.05% given  size  limitations  for  GET   0.02% requests,  the  POST  method  tends  to   0% 25.00% 50.00% 75.00% 100.00% be  reserved  for  situations  such  as   <illing  out  a  web  form  when  more   substantial  amounts  of  data  need  to   Transactions Request Size Response Size be  transmitted,  while  GET  requests   are  leveraged  for  general  web  page   rendering.  It  is,  however,   interesting  to  note  the  percentages  of  overall  traf<ic  related  to  request  and  response   size.  Even  though  POST  requests  accounted  for  7.18%  of  transactions  during  the   quarter,  (given  their  use  in  uploading  content),  such  requests  were  responsible  for  
  • 16. more  than  twice  as  much  (15.16%)  of  total  outbound  web  traf<ic.  HTTP  CONNECT   requests  by  contrast  have  the  opposite  effect.  Such  requests  are  used  to  initiate   traf<ic  on  an  alternate  port.  As  such  the  requests  are  limited  in  size.   Response  Code Top 10 Response Codes Q4 2009 0% 20.00% 40.00% 60.00% 80.00% 200 - OK 304 - Not Modified 8.94% 78.98% 302 - Found 3.22% 307 - Temporary Redirect 2.23% Invalid 2.00% 404 - Not Found 1.35% 204 - No Content 0.98% 403 - Forbidden 0.80% 301 - Moved Permanently 0.60% 206 - Partial Content 0.28% Just  over  4  out  of  5  requests  (80.29%)  returned  a  200  level  (success)  code   representing  that  the  content  had  been  delivered  and  no  further  action  was   required.  300  level  (redirection)  codes,  indicating    additional  action  required  by  the   requesting  browser  overall  accounted  for  15%  of  traf<ic.  Client  errors  (400)  were   relatively  rare  at  2.49%,  but  not  as  rare  as  server  errors  (500),  which  occurred  only   0.11%  of  the  time.  
  • 17. Web  Browser  Sta0s0cs Browser  Version Browser Market Share By Month Q4 2009 80.0% 60.0% 40.0% 20.0% IE Firefox Safari 0% Opera Chrome Unknown Other October November December While  Internet  Explorer  clearly  continues  to  dominate,  we  are  witnessing  a  slow  but   steady  decline  in  overall  market  share.  Regardless,  other  browser  vendors  have  a   long  way  to  go  before  they  will  surpass  the  long  standing  market  leader.  We  saw  a   greater  than  6%  jump  in  market  share  for  Firefox,  during  the  month  of  December.   However  this  can  be  largely  attributed  to  improved  detection  methods  as  opposed  to   an  unexpected  surge  in  traf<ic.  You  will  note  that  Unknown  traf<ic  declined  a  similar   amount  during  the  same  time  period.  Unknown  traf<ic  accounts  for  a  reasonable   amount  of  traf<ic  as  today  -­‐  the  majority  of  desktop  applications  communicate  via   HTTP/HTTPS  for  a  variety  of  reasons  including  the  retrieval  of  additional  content,   providing  online  support,  downloading  patches  and  submitting  error  reports.  Safari,   Opera  and  Chrome  combined,  continue  to  account  for  less  than  two  percent  of  the   traf<ic  that  we’re  seeing.  It  will  be  interesting  to  watch  Chrome  in  the  coming  months   as  Google  is  starting  to  leverage  its  reach  to  promote  the  browser.
  • 18. Internet Explorer Breakdown Q4 2009 Looking  at  the  breakdown  of   Internet  Explorer  traf<ic  for  the   quarter  is  particularly   concerning.  The  majority  of   5% enterprises  continue  to   maintain  Internet  Explorer  6.x   as  their  browser  of  choice.   1% While  IE  6  continues  to  be   supported  by  Microsoft,   meaning  that  patches  are   deployed  for  any  known   vulnerabilities,  it  lacks   48% numerous  security  features   46% present  in  IE  7  and  8.  IE  6  does   not  maintain  malicious  URL  and   phishing  block  lists,  a  feature   that  is  now  common  place  in  all   major  browsers  and  is  even   making  its  way  into  mobile   browsers.  Additionally,  IE  6   lacks  protections  such  as  Data   Execution  Prevention  (DEP)  and   IE 6.0 IE 7.0 Address  Space  Layout  Randomization   IE 8.0 IE Other (ASLR),  two  features  which  increase  the   complexity  of  executing  shellcode  should  a   remote  browser  exploit  be  uncovered.  During  the  Operation  Aurora  attack,  when   Google,  Adobe  and  other  high  pro<ile  enterprises  were  allegedly  in<iltrated  by   Chinese  attackers,  the  attacks  only  targeted  IE6.  While  IE  7  &  8  were  vulnerable  to   the  same  attack  vector,  reliable  exploit  code  had  not  been  produced  for  these   versions  on  the  browser  due  to  additional  protections  such  as  DEP  and  ASLR.  IE  8   also  added  a  critical  feature  which  has  now  been  adopted  by  chrome  -­‐  the  inclusion   of  cross-­‐site  scripting  protection,  yet  another  feature  that  IE  6  lacks.  It  is  vital  that   enterprises  move  away  from  IE  6,  even  though  it  continues  to  be  supported  by   Microsoft  and  adopt  IE  8  to  take  advantage  of  numerous  security  enhancements.   Google  has  indirectly  taken  an  important  step  toward  forcing  this  change  by   dropping  support  for  Google  Docs  and  Google  Sites  in  IE  6,  starting  in  March  2010. October November December Internet Explorer 75.31% 73.77% 72.21% Firefox 8.44% 8.87% 15.32% Safari 1.41% 1.38% 1.39% Opera 0.02% 0.06% 0.09% Chrome 0.06% 0.02% 0.03% Unknown 13.04% 14.18% 9.33% Other 1.72% 1.72% 1.63% 100.00% 100.00% 100.00%
  • 19. User  Sta0s0cs Attackers  are  no  longer  targeting  web  and  email  servers.  Instead,  they  are   focusing  on  the  weakest  link  in  the  security  chain  -­‐  end  users.  Whether  such   attacks  leverage  technical  vulnerabilities,  or  more  likely,  social  engineering   attacks,  web  based,  client-­‐side  attacks  are  the  most  common  way  to   compromise  end  user  machines.  As  such  it’s  vital  for  enterprises  to   understand  user  behavior  on  the  web. URL  Categoriza0on Given  the  corporate  focus  of  Zscaler  clients  it  isn’t  surprising  that  categories  such  as   Professional  Services  and  Corporate  Marketing  would  top  the  list.  More  interesting   are  the  high  placements  of  personal  traf<ic  such  as  Shopping,  Sports,  Entertainment   and  games.  In  fact,  the  majority  of  sites  beyond  the  top  10  are  personal  in  nature.   Overall,  approximately  1/5  of  traf<ic  could  be  deemed  to  be  personal  in  nature.   While  Zscaler  delivers  an  enterprise  offering  it  is  not  uncommon  for  employees  to   leverage  corporate  assets  after  work  hours  for  personal  purposes  and  as  such,  some   of  this  traf<ic  was  likely  generated  outside  of  work  hours. Below  we  breakdown  a  select  number  of  individual  categories  to  reveal  the  top  10   domains  within  each. Search  Engines The  search  engine  game  remains  a  three  horse   Top Search Engines race,  with  Google  continuing  to  dominate  the   majority  of  traf<ic.  After  the  big  three,   contenders  are  hard  to  <ind.  Disney’s  Go.com   Other which  is  actually  powered  by  Yahoo!,  sat  in  4th   15% place  at  1.22%  and  Baidu,  a  powerhouse  in  the   Chinese  market  handled  1.00%  of  web  search   Microsoft traf<ic  for  Q4  2009. 10% Yahoo! Google 18% 57%
  • 20. Social  networking The  dominance  of  Facebook  in   Top Social Networking Sites Q4 2009 the  social  networking  realm  is   clear.  Three  quarters  of  all  social   networking  traf<ic  traversing  the   Zscaler  network  is  destined  for   Other Facebook.  MySpace  has  solid   11% control  of  second  place  with   15%  of  traf<ic  but  the  gap   between  <irst  and  second  place   Myspace is  enormous  and  only  appears  to   15% be  getting  larger. It  is  also  interesting  to  consider   that  the  majority  of  traf<ic  in   these  statistics  is  corporate   traf<ic.  While  a  portion  of   requests  are  no  doubt  personal   Facebook in  nature,  this  also  suggests  that   74% Facebook  is  becoming  a  social   platform  of  choice  for   enterprises.  More  and  more,   corporations  are  attempting  to   leverage  social  networks  for  marketing,  recruiting  and  investigating  potential  new   hires. File  Sharing Si@mBIT,  a  Thailand  based  web   Top File Sharing Domains hosting  provider  describing   itself  as  “the  best  Thailand   Bittorrent  website  since  2005”   0% 10.00% 20.00% 30.00% led  statistics  for  the  quarter   40.00% with  37.49%  of  all  <ile  sharing   siambit.com 37.49% traf<ic.  The  third  largest   domain,  tb.in.th  is  also   filestube.com 25.31% controlled  by  Si@mBIT    and   tb.in.th 12.98% commanded  12.98%  of  traf<ic,   sftcdn.net 9.68% giving  Si@mBIT  approximately   iptorrents.com 3.06% half  of  the  traf<ic  for  the  quarter.   limewire.com 2.95% FilesTube,  a  search  engine   Other 1.98% dedicated  to  <ile  downloads  had   seedpeer.com 1.07% 25.31%  of  traf<ic.  
  • 21. Government Q4  means  that  Christmas  is  on   Top 10 Government Domains the  way  and  it  would  appear   that  the  United  States  Postal   Service  (USPS)  was  a  popular   0% 5.00% 10.00% 15.00% destination  for  holiday   20.00% shoppers  looking  to  determine   if  their  gifts  would  arrive  on   usps.com 16.20% time.  USPS  accounted  for   nraila.org 5.14% 16.20%  of  government  related   weather.gov 3.60% traf<ic  during  the  quarter.   uspto.gov 2.25% www.sec.gov 1.73% state.fl.us 1.47% fema.gov 1.38% www.irs.gov 1.36% michigan.gov 1.12% military.com 1.03% Retail Q4  is  of  course  the  peak  online   Top 10 Shopping Sites shopping  season,  a  time  when   retailers  look  to  make  the   majority  of  their  pro<it  for  the   0% 1.00% 2.00% 3.00% 4.00% year.  If  web  traf<ic  is  any   indication,  Amazon  was  the   Amazon 3.63% big  winner,  having  claimed   ShopLocal.com 2.90% 3.63%  of  total  retail  traf<ic.   Macy’s 2.59% ShopLocal,  which  took  the   Shop.com 2.55% Overstock 1.88% number  two  spot,  is  not  a   JC Penny 1.87% retailer  itself  but  rather  a  site   Target 1.52% which  republishes  <lyers  for   Costco 1.27% local  stores  to  allow  user  to   Barnes & Noble 1.10% QVC 1.10% <ind  deals  speci<ic  to  their   geographic  area.  The  company   makes  money  through   advertising  on  the  site.
  • 22. Security  Sta0s0cs Threats Next,  we’ll  breakdown  the  various  threats  that  we  see  on  a  daily  basis.  These  results   are  based  on  actual  end  user  traf<ic  and  therefore  re<lect  popular  and  active   malicious  sites  as  opposed  to  sites  that  may  exist  but  not  be  visited. Malware  By  IP  Address Worms,  viruses,  Trojans  and   Top 10 Malware IP Addresses other  forms  of  malware  can  be   found  just  about  everywhere  on   the  web  today.  However   0% 10.00%20.00% 30.00% 40.00% malicious  content  is  not   necessarily  hosted  at  sites  that   38.99.186.14 38.63% are  themselves  malicious.  More   208.71.120.24 25.68% and  more,  we’re  seeing  otherwise   208.71.121.24 13.41% legitimate  sites  hosting  malware   124.153.77.48 5.52% without  them  being  aware  of  it.   217.23.7.7 2.73% This  is  an  increasing  concern   64.14.29.50 1.32% given  the  trend  toward   216.86.150.237 1.01% 208.76.70.56 1.00% permitting  user  supplied  content   74.125.19.83 0.91% to  be  shared.  Unfortunately,  many   74.125.19.18 0.88% sites  are  doing  little  to  ensure   that  the  hosted  content  is  not   malicious  before  it  is  stored  for   others  to  access. Malware  by  Country Sites  hosted  in  the  United  States   overwhelmingly  hosted  the  majority  of   Top Countries Serving Malware malware  and  for  this  reason  we  have   broken  them  out  separately.  80.32%  of   malware  seen  during  Q4  2009  originated   from  US  based  servers.  This  should  not,   however,  be  interpreted  as  US-­‐based   Other traf<ic  being  particularly  risky,  rather  it’s   20% more  of  a  re<lection  of  the  fact  that  the   majority  of  traf<ic  inspected  was  destined   for  served  located  in  the  US.  This  can  be   seen  in  the  Geography  section  of  this   paper. United States 80%
  • 23. Top 10 Countries Serving Malware (US Excluded) 5% 3% 5% 25% 6% 6% 6% 11% 20% 14% Netherlands India Germany China Cyprus Russian Federation United Kingdom Canada Korea, Republic of France Phishing The  top  phishing  site  blocked   Top 10 Phishing IP Addresses was  coolxd.com  -­‐  this   accounted  for  roughly  70%  of   the  quarter's  phishing   80.00% numbers.  The  site  itself,  was   0% 20.00% 40.00% 60.00% recently  removed  from  the   208.43.210.147 70.83% Internet.  This  scam  site  is   219.232.243.74 7.21% effectively  the  same  as  the   219.232.243.65 1.91% heyxd.com,  omgxd.com,  and   219.232.243.91 1.64% 219.232.243.75 1.47% imnotez.com  sites.  These  sites   219.232.243.15 0.84% steal  your  email/instant   219.232.243.90 0.61% messenger  credentials   219.232.241.178 0.57% 219.232.243.87 0.55% (username/password),  and   174.143.29.2 0.50% then  noti<ies  the  people  on   your  contact  list  to  check  out   the  site.  Advertisements,   fraud,  and/or  malware  are   then  spammed  to  and  through   victim  accounts.  The  sites  advertised  the  ability  to  provide  a  service  which  enables   users  to  IM  pictures  and  other  content  to  share  directly  to  a  forum.
  • 24. Malicious  Domains Three  domains  accounted  for   Top 10 Malicious Domains roughly  55%  of  the  malicious   URLs  transactions: 0% 10.00% 20.00% 30.00% •adfarm.mediaplex.com •link4you.3322.org adfarm.mediaplex.com 24.01% •www.tns-­‐counter.ru link4you.3322.org 17.41% www.tns-counter.ru 13.33% adfarm.mediaplex.com,  has  been   www.winifixer.com 4.06% reported  to  be  involved  in  spam,   www.freegaming.de 2.96% adware/spyware,  phishing/ dt.tongji.linezing.com 2.25% img.12chan.org 1.72% scams,  and  browser  exploits5.   nspmotion.com 1.14% The  Mediaplex  website  details   acs86.com 0.69% how  the  company  "provides   stork27.dropbox.com 0.66% cross-­‐channel  advertising   technology  solutions  and   services  that  enable  marketers  to   achieve  one-­‐ Top Malicious Domains By Country United States to-­‐one   Canada messaging,   Russian Federation greater   China ef<iciencies   6% Germany and  a   3% Netherlands 5% competitive   Other edge  through   6% insightful   reporting  and  analytics” 6.  3322.org  is  a   DynDNS  provided  domain  that  has   44% served  malware  and  exploit  content  for   17% some  time7.  tns-­‐counter.ru  is  also   known  for  serving  adware/spyware/ malware 8. The  majority  of  malicious  sites  are   19% hosted  in  the  US,  with  a  full  63%  of  sites   residing  in  North  America.  This  is   however  more  a  re<lection  of  where   content  in  general  resides  as  opposed  to  North  American  content  representing  a   higher  overall  risk. 5 http://www.siteadvisor.com/sites/mediaplex.com/summary/ 6 http://www.mediaplex.com/about.shtml 7 http://isc.sans.org/diary.html?storyid=5710 8 http://www.siteadvisor.com/sites/tns-­‐counter.ru/summary/
  • 25. Anonymizers Top 10 Anonymizers Over  30%  of  our   anonymizer  traf<ic  was  to   kproxy.com.  One  of  the   features  that  Zscaler   0% 10.00% 20.00% 30.00% 40.00% provides  to  customers,  is   policy  based  blocking  based   kproxy.com 30.51% on  page  categorization.  So   proxyswitcher.com 20.17% customers  have  the  ability   freeproxylist.org 8.03% to  block  users  from   archive.org 5.36% browsing  to/through  proxy   freeproxy.ru 3.12% sites.  kproxy.com  provides   privacy-world.com 1.83% a  simple  interface,  not   helllabs.net 1.76% unlike  Google’s,  to  browse   66.232.118.93 1.66% proxybridge.com 1.57% through,  with  SSL   ktunnel.com 1.39% encryption  as  an  additional   capability.  Of  the  popular   sites  that  kproxy  advertises   that  it  works  with  are   MySpace,  Facebook,  Gmail,   YouTube,  and  MegaUpload  -­‐  all  sites,  that  may  be  blocked  by  company  policies  as   they  are  not  work  related.  In  other  words,  users  are  generally  using  these  services  to   get  around  corporate  policies  and  URL  <iltering  rules  as  opposed  to  using  them  to   cloak  their  IP  address  from  an  external  source. Botnets Generally  speaking,  by   Top 10 Botnets IPs/Domains correlating  the  malicious   artifact  to  the  top  botnet   hosts,  enables  us  to   0% 12.50% 25.00% 37.50% describe  which  malware   50.00% campaigns  were  the  most   91.212.65.13 44.11% successful.  The  breakdown   66.235.175.5 15.67% is  as  follows,  and  should   77.221.133.227 9.80% not  be  of  surprise  to  the   88.80.7.152 8.39% security  community  for   88.80.5.3 7.05% HTTP  based  botnets: 77.221.133.189 5.74% 208.99.193.130 3.18% 1.Zeus/Zbot  variants meu89.net 1.91% 2.Fake  Anti-­‐Virus  variants 194.68.45.50 1.63% 3.Banker  Trojan  variants. 69.61.21.115 0.28% The  top  command  and   control  IP  address  seen,   91.212.65.13,  is  based  out   of  the  Ukraine  and  serviced   both  Zeus  and  FakeAV  infections.  The  whois  information  for  this  host  shows  it  
  • 26. belonging  to  the  Eurohost/UralComp  IP  blocks.  FireEye  has  a  good  write-­‐up  of  this   "bad  actor"  from  almost  a  year  ago9  and  malwaredomainlist,  an  archive  of  malicious   web  domains  has  plenty  of  content  for  these  IP  blocks10 . While  Ukraine  and  Russian  IPs  make  up  a  large  number  of  the  botnet  C&C  servers,  it   was  a  little  surprising  to  see  that  Sweden  had  a  number  of  C&Cs  in  the  top  25: • 88.80.7.152 • 88.80.5.3 • 88.80.5.172 • 80.88.108.18 Further  analysis  of  some  of  the  Swedish  hosts  shows  them  belonging  to  PRQ  (http:// www.prq.se)  a  co-­‐location  and  hosting  provider.    Their  homepage  states  that  they   are  known  for  their  "boundless  commitment  to  free  speech"  and  "discrete  customer   relations  policy".  They  also  have  an  icon  on  their  website  that  states,  "data  retention   is  no  solution",  suggesting  minimal/no  logging.    In  other  words,  this  hosting  service   would  be  ideal  for  hosting  malicious  sites  and  remaining  protected  from   investigations  /  takedowns. Traffic Last,  but  not  least,  we’ll  investigate  traf<ic  patterns  which  would  not  be  expected   without  the  presence  of  errors  of  malicious  content. Bogon  IP  space Top 10 Bogon IP Addresses Bogon  (aka  darknet)  IP  addresses   represent  non-­‐routable  IP  blocks,   either  because  they  are  reserved   0% 2.00% 4.00% 6.00% 8.00% (for  example  RFC1918)  or  they   are  unallocated.  Occasionally,  we   1.1.1.1 7.74% 127.0.0.0 6.60% see  web  requests  to  bogon  IPs  -­‐   198.18.1.18 5.35% usually  this  is  to  RFC1918   1.2.3.4 4.34% 0.0.0.2 2.99% address  (internal  IP  addresses),   0.0.0.5 2.62% 198.18.1.15 2.60% and  the  requests  have  leaked  into   0.0.0.8 2.57% the  cloud  because  of  a  routing   198.18.1.2 2.20% 0.0.0.1 2.20% miscon<iguration  on  the   9 http://blog.fireeye.com/research/2009/03/bad-actors-part-6-eurohost-llc.html 10 http://www.malwaredomainlist.com/forums/index.php?board=23.0
  • 27. Top 10 Bogon IP Address Blocks customer's  network.  However,   there  are  also,  several   0% occurrences  of  web  requests  to   1.75% 3.50% 5.25% 7.00% non-­‐RFC1918  bogons.  This  traf<ic   127.0.0.0/24 6.60% is  of  interest  as  it  represents   1.2.3.0/24 4.34% either  human  error  or  an  infected   0.0.0.0/24 2.62% 50.0.0.0/24 2.04% machine  that  is  randomly   169.254.1.0/24 1.11% scanning  IP  address  blocks   169.254.178.0/24 0.56% 169.254.200.0/24 0.53% looking  for  vulnerable  hosts. 169.254.8.0/24 0.37% 198.18.189.0/24 0.37% 0.1.0.0/24 0.34% Some  of  the  bogon  traf<ic  can  be   explained  as  follows: • The  1.1.1.1  and  127.0.0.0/8  and  1.2.3.0/24  subnets  are  likely  some  sort  of  test   scripts  that  folks  are  running. • The  169.254.0.0/16  addresses  are  part  of  the  Automatic  Private  Addressing   (APIPA)  of  hosts  when  DHCP  fails. The  50.0.0.0/24  IP  block  is  interesting,  though  yet  unexplained.  Googling  for  it   shows  that  it  is  an  IANA  reserved  block,  and  it  shows  up  in  some  OSPF  routing   templates.  It's  possible  that  this  block  is  a  commonly  used  reserve  block  in  some   intra-­‐organization  routing.  However,  the  only  IP  address  that  was  hit  in  this  block   was  50.0.0.82,  which  is  interesting.  It  is  possible  that  there  was  a  mistake  in  a  script   or  routing  statement.
  • 28. Conclusion Understanding  web  traf<ic  is  critical  for  enterprises  seeking  to  manage  and  secure   their  networks.  Traf<ic  is  converging  on  the  web  at  a  rapid  pace.  A  decade  ago  we   leveraged  <irewalls  to  manage  traf<ic  on  networks  and  determine  which  users  could   access  which  resources.  Today,  traf<ic  is  not  neatly  segregated  into  buckets  based  on   protocols.  Regardless  of  the  traf<ic  that  we’re  dealing  with,  be  it  email,  instant   messaging,  P2P,  streaming  media,  etc.,  it  has  the  ability  to  be  tunneled  through   HTTP/HTTPS. At  the  same  time,  attackers  have  shifted  their  focus  to  target  end  users.  Some   attackers  take  a  shotgun  approach  by  striking  far  and  wide  without  concern  for  who   the  ultimate  victims  may  be.  This  is  the  approach  leveraged  by  those  who  build   botnets.  They  seek  infected  machines  and  they  do  not  discriminate.  On  the  other   side  of  the  coin,  Advanced  Persistent  Threats11  are  emerging  on  the  radars  of  CISOs   as  the  media  highlights  the  sophistication  of  attacks  on  corporations,  such  as  those   highlighted  in  the  Operation  Aurora  attacks  which  targeted  Google,  Adobe  and   others.  Regardless  of  the  approach,  the  majority  of  such  attacks  now  leverage  the   web  as  the  transport  medium.   Understanding  the  behaviors  of  end  users,  content  providers  and  attackers  on  the   web  can  help  us  to  better  manage  and  secure  networks.  We  hope  that  you  enjoyed   this,  our  <irst  quarterly  State  of  the  Web  report.   11 http://www.zscaler.com/apt.html
  • 29. Appendix TLD  Breakdown Monthly  Summary  –  Top  TLDs  Visited Note: Pink shows larger fluctuations than yellow, and green shows no fluctuation Monthly Summary – Top TLDs by Transactions Popularity October 2009 November 2009 December 2009 1 COM COM COM 2 NET NET NET 3 ORG ORG ORG 4 AU AU AU 5 IN UK TV 6 TV IN ZA 7 UK TV IN 8 FR DE UK 9 PE PE DE 10 GOV FR GOV 11 EDU ZA EDU 12 DE GOV RU 13 RU NU FR 14 US EDU US 15 NU RU PE 16 IT IT CN 17 AR US IT 18 MX MX CA 19 CA CN SG 20 INFO AR MX 21 CO CA INFO 22 BR IE NU 23 CN INFO AR 24 ES TH PL 25 FM ES FM