SlideShare ist ein Scribd-Unternehmen logo
1 von 36
Downloaden Sie, um offline zu lesen
How	
  to	
  get	
  the	
  MTTR	
  below	
  1	
  
minute	
  and	
  more	
  
Devaraj	
  Das	
  
(ddas@hortonworks.com)	
  
Nicolas	
  Liochon	
  
(nkeywal@gmail.com)	
  
Outline	
  
•  What	
  is	
  this?	
  Why	
  are	
  we	
  talking	
  about	
  this	
  
topic?	
  Why	
  it	
  ma>ers?	
  ….	
  
•  HBase	
  Recovery	
  –	
  an	
  overview	
  
•  HDFS	
  issues	
  
•  Beyond	
  MTTR	
  (Performance	
  post	
  recovery)	
  
•  Conclusion	
  /	
  Future	
  /	
  Q	
  &	
  A	
  
What	
  is	
  MTTR?	
  Why	
  its	
  important?	
  …	
  
•  Mean	
  Time	
  To	
  Recovery	
  -­‐>	
  Average	
  Pme	
  required	
  
to	
  repair	
  a	
  failed	
  component	
  (Courtesy:	
  Wikipedia)	
  
•  Enterprises	
  want	
  an	
  MTTR	
  of	
  ZERO	
  
–  Data	
  should	
  always	
  be	
  available	
  with	
  no	
  degradaPon	
  
of	
  perceived	
  SLAs	
  
–  PracPcally	
  hard	
  to	
  obtain	
  but	
  yeah	
  it’s	
  a	
  goal	
  
•  Close	
  to	
  Zero-­‐MTTR	
  is	
  especially	
  important	
  for	
  
HBase	
  
–  Given	
  it	
  is	
  used	
  in	
  near	
  realPme	
  systems	
  
•  MTTR	
  in	
  other	
  NoSQL	
  systems	
  &	
  Databases	
  
HBase	
  Basics	
  
•  Strongly	
  consistent	
  
–  Write	
  ordered	
  with	
  reads	
  
–  Once	
  wri>en,	
  the	
  data	
  will	
  stay	
  
•  Built	
  on	
  top	
  of	
  HDFS	
  
•  When	
  a	
  machine	
  fails	
  the	
  cluster	
  remains	
  
available,	
  and	
  its	
  data	
  as	
  well	
  
•  We’re	
  just	
  speaking	
  about	
  the	
  piece	
  of	
  data	
  that	
  
was	
  handled	
  by	
  this	
  machine	
  
Write	
  path	
  
WAL	
  –	
  Write	
  
Ahead	
  Log	
  
A	
  write	
  is	
  
finished	
  once	
  
wri>en	
  on	
  all	
  
HDFS	
  nodes	
  
The	
  client	
  
communicated	
  
with	
  the	
  region	
  
servers	
  
We’re	
  in	
  a	
  distributed	
  system	
  
•  You	
  can’t	
  disPnguish	
  a	
  
slow	
  server	
  from	
  a	
  
dead	
  server	
  
•  Everything,	
  or,	
  nearly	
  
everything,	
  is	
  based	
  
on	
  Pmeout	
  
•  Smaller	
  Pmeouts	
  means	
  more	
  false	
  posiPve	
  
•  HBase	
  works	
  well	
  with	
  false	
  posiPve,	
  but	
  they	
  
always	
  have	
  a	
  cost.	
  
•  The	
  less	
  the	
  Pmeouts	
  the	
  be>er	
  
HBase	
  components	
  for	
  recovery	
  
Recovery	
  in	
  acPon	
  
Recovery	
  process	
  
•  Failure	
  detecPon:	
  ZooKeeper	
  
heartbeats	
  the	
  servers.	
  Expire	
  
the	
  session	
  when	
  it	
  does	
  not	
  
reply	
  
•  Region	
  assignment:	
  the	
  master	
  
reallocates	
  the	
  regions	
  to	
  the	
  
other	
  servers	
  
•  Failure	
  recovery:	
  read	
  the	
  WAL	
  
and	
  rewrite	
  the	
  data	
  again	
  
•  The	
  clients	
  stops	
  the	
  
connecPon	
  to	
  the	
  dead	
  server	
  
and	
  goes	
  to	
  the	
  new	
  one.	
  
ZK	
  
Heartbeat	
  
Client	
  
Region	
  Servers,	
  
DataNode	
  
Data	
  recovery	
  
Master,	
  RS,	
  ZK	
  
Region	
  Assignment	
  
So….	
  
•  Detect	
  the	
  failure	
  as	
  fast	
  as	
  possible	
  
•  Reassign	
  as	
  fast	
  as	
  possible	
  
•  Read	
  /	
  rewrite	
  the	
  WAL	
  as	
  fast	
  as	
  possible	
  
•  That’s	
  obvious	
  
The	
  obvious	
  –	
  failure	
  detecPon	
  
•  Failure	
  detecPon	
  
–  Set	
  a	
  ZooKeeper	
  Pmeout	
  to	
  30s	
  instead	
  of	
  the	
  old	
  180s	
  
default.	
  	
  
–  Beware	
  of	
  the	
  GC,	
  but	
  lower	
  values	
  are	
  possible.	
  
–  ZooKeeper	
  detects	
  the	
  errors	
  sooner	
  than	
  the	
  configured	
  
Pmeout	
  
•  0.96	
  	
  
–  HBase	
  scripts	
  clean	
  the	
  ZK	
  node	
  when	
  the	
  server	
  is	
  kill	
  
-­‐9ed	
  
•  =>	
  DetecPon	
  Pme	
  becomes	
  0	
  
–  Can	
  be	
  used	
  by	
  any	
  monitoring	
  tool	
  
The	
  obvious	
  –	
  faster	
  data	
  recovery	
  
•  Not	
  so	
  obvious	
  actually	
  
•  Already	
  distributed	
  since	
  0.92	
  
–  The	
  large	
  the	
  cluster	
  the	
  be>er.	
  
•  Completely	
  rewri>en	
  in	
  0.96	
  
–  Recovery	
  itself	
  rewri>en	
  in	
  0.96	
  
–  Will	
  be	
  covered	
  in	
  the	
  second	
  part	
  
The	
  obvious	
  –	
  Faster	
  assignment	
  
•  Faster	
  assignment	
  
–  Just	
  improving	
  performances	
  
•  Parallelism	
  
•  Speed	
  
–  Globally	
  ‘much’	
  faster	
  
–  Backported	
  to	
  0.94	
  
•  SPll	
  possible	
  to	
  do	
  be>er	
  for	
  huge	
  number	
  of	
  
regions.	
  	
  
•  A	
  few	
  seconds	
  for	
  most	
  cases	
  
With	
  this	
  
•  DetecPon:	
  from	
  180s	
  to	
  30s	
  
•  Data	
  recovery:	
  around	
  10s	
  
•  Reassignment	
  :	
  from	
  10s	
  of	
  seconds	
  to	
  
seconds	
  
Do	
  you	
  think	
  we’re	
  be>er	
  with	
  this	
  
•  Answer	
  is	
  NO	
  
•  Actually,	
  yes	
  but	
  if	
  and	
  only	
  if	
  HDFS	
  is	
  fine	
  
– But	
  when	
  you	
  lose	
  a	
  regionserver,	
  you’ve	
  just	
  lost	
  
a	
  datanode	
  
DataNode	
  crash	
  is	
  expensive!	
  
•  One	
  replica	
  of	
  WAL	
  edits	
  is	
  on	
  the	
  crashed	
  DN	
  
– 33%	
  of	
  the	
  reads	
  during	
  the	
  regionserver	
  recovery	
  
will	
  go	
  to	
  it	
  
•  Many	
  writes	
  will	
  go	
  to	
  it	
  as	
  well	
  (the	
  smaller	
  
the	
  cluster,	
  the	
  higher	
  that	
  probability)	
  
•  NameNode	
  re-­‐replicates	
  the	
  data	
  (maybe	
  TBs)	
  
that	
  was	
  on	
  this	
  node	
  to	
  restore	
  replica	
  count	
  
– NameNode	
  does	
  this	
  work	
  only	
  amer	
  a	
  good	
  
Pmeout	
  (10	
  minutes	
  by	
  default)	
  
HDFS	
  –	
  Stale	
  mode	
  
Live	
  
Stale	
  
Dead	
  
As	
  today:	
  used	
  for	
  reads	
  &	
  
writes,	
  using	
  locality	
  
Not	
  used	
  for	
  writes,	
  used	
  as	
  
last	
  resort	
  for	
  reads	
  
As	
  today:	
  not	
  used.	
  
And	
  actually,	
  it’s	
  be>er	
  to	
  do	
  the	
  HBase	
  
recovery	
  before	
  HDFS	
  replicates	
  the	
  TBs	
  
of	
  data	
  of	
  this	
  node	
  
30	
  seconds,	
  can	
  be	
  less.	
  
10	
  minutes,	
  don’t	
  change	
  this	
  
Results	
  
•  Do	
  more	
  read/writes	
  to	
  HDFS	
  during	
  the	
  
recovery	
  
•  MulPple	
  failures	
  are	
  sPll	
  possible	
  
– Stale	
  mode	
  will	
  sPll	
  play	
  its	
  role	
  
– And	
  set	
  dfs.Pmeout	
  to	
  30s	
  
– This	
  limits	
  the	
  effect	
  of	
  two	
  failure	
  in	
  a	
  row.	
  The	
  
cost	
  of	
  the	
  second	
  failure	
  is	
  30s	
  if	
  you	
  were	
  
unlucky	
  
Are	
  we	
  done?	
  
•  We’re	
  not	
  bad	
  
•  But	
  there	
  is	
  sPll	
  something	
  
The	
  client	
  
You	
  lem	
  it	
  waiPng	
  on	
  the	
  dead	
  server	
  
	
  	
  	
  	
  
Here	
  it	
  is	
  
The	
  client	
  
•  You	
  want	
  the	
  client	
  to	
  be	
  paPent	
  
•  Retries	
  when	
  the	
  system	
  is	
  already	
  loaded	
  is	
  
not	
  good.	
  	
  
•  You	
  want	
  the	
  client	
  to	
  learn	
  about	
  region	
  
servers	
  dying,	
  and	
  to	
  be	
  able	
  to	
  react	
  
immediately.	
  
•  You	
  want	
  this	
  to	
  scale.	
  
SoluPon	
  
•  The	
  master	
  noPfies	
  the	
  client	
  
–  A	
  cheap	
  mulPcast	
  message	
  with	
  the	
  “dead	
  servers”	
  
list.	
  Sent	
  5	
  Pmes	
  for	
  safety.	
  
–  Off	
  by	
  default.	
  
–  On	
  recepPon,	
  the	
  client	
  stops	
  immediately	
  waiPng	
  on	
  
the	
  TCP	
  connecPon.	
  You	
  can	
  now	
  enjoy	
  large	
  
hbase.rpc.Pmeout	
  
Full	
  workflow	
  
t0	
  
t1	
  
t2	
  
t3	
  
Client	
  reads	
  
and	
  writes	
  
RegionServer	
  serving	
  
reads	
  and	
  writes	
  
RegionServer	
  crashes	
  
Affected	
  regions	
  
reassigned	
  
Client	
  writes	
  
Data	
  recovered	
  
Client	
  reads	
  
and	
  writes	
  t4	
  
Are	
  we	
  done	
  
•  In	
  a	
  way,	
  yes	
  
– There	
  is	
  a	
  lot	
  of	
  things	
  around	
  asynchronous	
  
writes,	
  reads	
  during	
  recovery	
  
– Will	
  be	
  for	
  another	
  Pme,	
  but	
  there	
  will	
  be	
  some	
  
nice	
  things	
  in	
  0.96	
  
•  And	
  a	
  couple	
  of	
  them	
  is	
  presented	
  in	
  the	
  
second	
  part	
  of	
  this	
  talk!	
  
Faster	
  recovery	
  
•  Previous	
  algo	
  
–  Read	
  the	
  WAL	
  files	
  
–  Write	
  new	
  Hfiles	
  
–  Tell	
  the	
  region	
  server	
  it	
  got	
  new	
  Hfiles	
  
•  Put	
  pressure	
  on	
  namenode	
  
–  Remember:	
  don’t	
  put	
  pressure	
  on	
  the	
  namenode	
  
•  New	
  algo:	
  
–  Read	
  the	
  WAL	
  
–  Write	
  to	
  the	
  regionserver	
  
–  We’re	
  done	
  (have	
  seen	
  great	
  improvements	
  in	
  our	
  tests)	
  
–  TBD:	
  Assign	
  the	
  WAL	
  to	
  a	
  RegionServer	
  local	
  to	
  a	
  replica	
  
RegionServer0	
   RegionServer_x	
  
RegionServer_y	
  
WAL-­‐file3	
  
<region2:edit1><region1:edit2>	
  
……	
  
<region3:edit1>	
  
……..	
  
WAL-­‐file2	
  
<region2:edit1><region1:edit2>	
  
……	
  
<region3:edit1>	
  
……..	
  
WAL-­‐file1	
  
<region2:edit1><region1:edit2>	
  
……	
  
<region3:edit1>	
  
……..	
  
HDFS	
  
Splitlog-­‐file-­‐for-­‐region3	
  
<region3:edit1><region1:edit2>	
  
……	
  
<region3:edit1>	
  
……..	
  
Splitlog-­‐file-­‐for-­‐region2	
  
<region2:edit1><region1:edit2>	
  
……	
  
<region2:edit1>	
  
……..	
  
Splitlog-­‐file-­‐for-­‐region1	
  
<region1:edit1><region1:edit2>	
  
……	
  
<region1:edit1>	
  
……..	
  
HDFS	
  
RegionServer3	
  
RegionServer2	
  
RegionServer1	
  
writes	
  
writes	
  
reads	
  
reads	
  
Distributed	
  log	
  
Split	
  
RegionServer0	
   RegionServer_x	
  
RegionServer_y	
  
WAL-­‐file3	
  
<region2:edit1><region1:edit2>	
  
……	
  
<region3:edit1>	
  
……..	
  
WAL-­‐file2	
  
<region2:edit1><region1:edit2>	
  
……	
  
<region3:edit1>	
  
……..	
  
WAL-­‐file1	
  
<region2:edit1><region1:edit2>	
  
……	
  
<region3:edit1>	
  
……..	
  
HDFS	
  
Recovered-­‐file-­‐for-­‐region3	
  
<region3:edit1><region1:edit2>	
  
……	
  
<region3:edit1>	
  
……..	
  
Recovered-­‐file-­‐for-­‐region2	
  
<region2:edit1><region1:edit2>	
  
……	
  
<region2:edit1>	
  
……..	
  
Recovered-­‐file-­‐for-­‐region1	
  
<region1:edit1><region1:edit2>	
  
……	
  
<region1:edit1>	
  
……..	
  
HDFS	
  
RegionServer3	
  
RegionServer2	
  
RegionServer1	
  
writes	
  
writes	
  
reads	
  
reads	
  
Distributed	
  log	
  
Replay	
  
replays	
  
Write	
  during	
  recovery	
  
•  Hey,	
  you	
  can	
  write	
  during	
  the	
  WAL	
  replay	
  
•  Events	
  stream:	
  your	
  new	
  recovery	
  Pme	
  is	
  the	
  
failure	
  detecPon	
  Pme:	
  max	
  30s,	
  likely	
  less!	
  
MemStore	
  flush	
  
•  Real	
  life:	
  some	
  tables	
  are	
  updated	
  at	
  a	
  given	
  
moment	
  then	
  lem	
  alone	
  
– With	
  a	
  non	
  empty	
  memstore	
  
– More	
  data	
  to	
  recover	
  
•  It’s	
  now	
  possible	
  to	
  guarantee	
  that	
  we	
  don’t	
  
have	
  MemStore	
  with	
  old	
  data	
  
•  Improves	
  real	
  life	
  MTTR	
  
•  Helps	
  snapshots	
  
.META.	
  
•  .META.	
  
–  There	
  is	
  no	
  –ROOT-­‐	
  in	
  0.95/0.96	
  
–  But	
  .META.	
  failures	
  are	
  criPcal	
  
•  A	
  lot	
  of	
  small	
  improvements	
  
–  Server	
  now	
  says	
  to	
  the	
  client	
  when	
  a	
  region	
  has	
  
moved	
  (client	
  can	
  avoid	
  going	
  to	
  meta)	
  
•  And	
  a	
  big	
  one	
  
–  .META.	
  WAL	
  is	
  managed	
  separately	
  to	
  allow	
  an	
  
immediate	
  recovery	
  of	
  META	
  
–  With	
  the	
  new	
  MemStore	
  flush,	
  ensure	
  a	
  quick	
  
recovery	
  
Data	
  locality	
  post	
  recovery	
  
•  HBase	
  performance	
  depends	
  on	
  data-­‐locality	
  
•  Amer	
  a	
  recovery,	
  you’ve	
  lost	
  it	
  
–  Bad	
  for	
  performance	
  
•  Here	
  comes	
  region	
  groups	
  
•  Assign	
  3	
  favored	
  RegionServers	
  for	
  every	
  region	
  
•  On	
  failures	
  assign	
  the	
  region	
  to	
  one	
  of	
  the	
  
secondaries	
  
•  The	
  data-­‐locality	
  issue	
  is	
  minimized	
  on	
  failures	
  
Block1	
   Block2	
   Block3	
  
Block1	
   Block2	
  
Rack1	
  
Block3	
  
Block3	
  
Rack2	
   Rack3	
  
Block1	
   Block2	
  
Datanode	
  
RegionServer1	
  
Datanode1	
  
RegionServer1	
  
Datanode	
  
RegionServer2	
  
Datanode1	
  
RegionServer1	
  
Datanode	
  
RegionServer3	
  
Block1	
   Block2	
  
Rack1	
  
Block3	
  
Block3	
  
Rack2	
   Rack3	
  
Block1	
   Block2	
  
RegionServer4	
   Datanode1	
  
RegionServer1	
  
Datanode	
  
RegionServer2	
  
Datanode1	
  
RegionServer1	
  
Datanode	
  
RegionServer3	
  
Reads	
  Blk1	
  and	
  
Blk2	
  remotely	
  
Reads	
  Blk3	
  
remotely	
  
RegionServer1	
  serves	
  three	
  regions,	
  and	
  their	
  StoreFile	
  blks	
  are	
  sca>ered	
  
across	
  the	
  cluster	
  with	
  one	
  replica	
  local	
  to	
  RegionServer1.	
  
Block1	
   Block2	
   Block3	
  
Block1	
   Block2	
  
Rack1	
  
Block3	
  
Block3	
  
Rack2	
   Rack3	
  
Block1	
   Block2	
  
Datanode	
  
RegionServer1	
  
Datanode1	
  
RegionServer1	
  
Datanode	
  
RegionServer2	
  
Datanode1	
  
RegionServer1	
  
Datanode	
  
RegionServer3	
  
RegionServer1	
  serves	
  three	
  regions,	
  and	
  their	
  StoreFile	
  blks	
  are	
  placed	
  on	
  
specific	
  machines	
  on	
  the	
  other	
  racks	
  
Block1	
   Block2	
  
Rack1	
  
Block3	
  
Block3	
  
Rack2	
   Rack3	
  
Block1	
   Block2	
  
RegionServer4	
   Datanode1	
  
RegionServer1	
  
Datanode	
  
RegionServer2	
  
Datanode1	
  
RegionServer1	
  
Datanode	
  
RegionServer3	
  
No	
  remote	
  reads	
  
Datanode	
  
Conclusion	
  
•  The	
  target	
  was	
  “from	
  omen	
  10	
  minutes	
  to	
  
always	
  less	
  than	
  1	
  minute”	
  
– We’re	
  almost	
  there	
  
•  Most	
  of	
  it	
  is	
  available	
  in	
  0.96,	
  some	
  parts	
  were	
  
backported	
  
•  Real	
  life	
  tesPng	
  of	
  the	
  improvements	
  in	
  
progress	
  
•  Room	
  for	
  more	
  improvements	
  
Q	
  &	
  A	
  
Thanks!	
  

Weitere ähnliche Inhalte

Was ist angesagt?

From A to Z-itrix: Setting up the most stable and fastest HCL Notes client on...
From A to Z-itrix: Setting up the most stable and fastest HCL Notes client on...From A to Z-itrix: Setting up the most stable and fastest HCL Notes client on...
From A to Z-itrix: Setting up the most stable and fastest HCL Notes client on...panagenda
 
Gpdb best practices v a01 20150313
Gpdb best practices v a01 20150313Gpdb best practices v a01 20150313
Gpdb best practices v a01 20150313Sanghee Lee
 
MongoDB and Azure Databricks
MongoDB and Azure DatabricksMongoDB and Azure Databricks
MongoDB and Azure DatabricksMongoDB
 
Tuning Apache Kafka Connectors for Flink.pptx
Tuning Apache Kafka Connectors for Flink.pptxTuning Apache Kafka Connectors for Flink.pptx
Tuning Apache Kafka Connectors for Flink.pptxFlink Forward
 
Best Practices with PostgreSQL on Solaris
Best Practices with PostgreSQL on SolarisBest Practices with PostgreSQL on Solaris
Best Practices with PostgreSQL on SolarisJignesh Shah
 
What Is Hadoop | Hadoop Tutorial For Beginners | Edureka
What Is Hadoop | Hadoop Tutorial For Beginners | EdurekaWhat Is Hadoop | Hadoop Tutorial For Beginners | Edureka
What Is Hadoop | Hadoop Tutorial For Beginners | EdurekaEdureka!
 
Flink Forward Berlin 2018: Stefan Richter - "Tuning Flink for Robustness and ...
Flink Forward Berlin 2018: Stefan Richter - "Tuning Flink for Robustness and ...Flink Forward Berlin 2018: Stefan Richter - "Tuning Flink for Robustness and ...
Flink Forward Berlin 2018: Stefan Richter - "Tuning Flink for Robustness and ...Flink Forward
 
Maximize Greenplum For Any Use Cases Decoupling Compute and Storage - Greenpl...
Maximize Greenplum For Any Use Cases Decoupling Compute and Storage - Greenpl...Maximize Greenplum For Any Use Cases Decoupling Compute and Storage - Greenpl...
Maximize Greenplum For Any Use Cases Decoupling Compute and Storage - Greenpl...VMware Tanzu
 
Batch Processing at Scale with Flink & Iceberg
Batch Processing at Scale with Flink & IcebergBatch Processing at Scale with Flink & Iceberg
Batch Processing at Scale with Flink & IcebergFlink Forward
 
The Columnar Era: Leveraging Parquet, Arrow and Kudu for High-Performance Ana...
The Columnar Era: Leveraging Parquet, Arrow and Kudu for High-Performance Ana...The Columnar Era: Leveraging Parquet, Arrow and Kudu for High-Performance Ana...
The Columnar Era: Leveraging Parquet, Arrow and Kudu for High-Performance Ana...DataWorks Summit/Hadoop Summit
 
Enabling the Active Data Warehouse with Apache Kudu
Enabling the Active Data Warehouse with Apache KuduEnabling the Active Data Warehouse with Apache Kudu
Enabling the Active Data Warehouse with Apache KuduGrant Henke
 
Big Data and Hadoop
Big Data and HadoopBig Data and Hadoop
Big Data and HadoopFlavio Vit
 
Room 3 - 7 - Nguyễn Như Phúc Huy - Vitastor: a fast and simple Ceph-like bloc...
Room 3 - 7 - Nguyễn Như Phúc Huy - Vitastor: a fast and simple Ceph-like bloc...Room 3 - 7 - Nguyễn Như Phúc Huy - Vitastor: a fast and simple Ceph-like bloc...
Room 3 - 7 - Nguyễn Như Phúc Huy - Vitastor: a fast and simple Ceph-like bloc...Vietnam Open Infrastructure User Group
 
Where is my bottleneck? Performance troubleshooting in Flink
Where is my bottleneck? Performance troubleshooting in FlinkWhere is my bottleneck? Performance troubleshooting in Flink
Where is my bottleneck? Performance troubleshooting in FlinkFlink Forward
 
Introduction to HADOOP.pdf
Introduction to HADOOP.pdfIntroduction to HADOOP.pdf
Introduction to HADOOP.pdf8840VinayShelke
 
Running Apache NiFi with Apache Spark : Integration Options
Running Apache NiFi with Apache Spark : Integration OptionsRunning Apache NiFi with Apache Spark : Integration Options
Running Apache NiFi with Apache Spark : Integration OptionsTimothy Spann
 
Hive 3 - a new horizon
Hive 3 - a new horizonHive 3 - a new horizon
Hive 3 - a new horizonThejas Nair
 
HBaseConAsia2018 Keynote 2: Recent Development of HBase in Alibaba and Cloud
HBaseConAsia2018 Keynote 2: Recent Development of HBase in Alibaba and CloudHBaseConAsia2018 Keynote 2: Recent Development of HBase in Alibaba and Cloud
HBaseConAsia2018 Keynote 2: Recent Development of HBase in Alibaba and CloudMichael Stack
 

Was ist angesagt? (20)

From A to Z-itrix: Setting up the most stable and fastest HCL Notes client on...
From A to Z-itrix: Setting up the most stable and fastest HCL Notes client on...From A to Z-itrix: Setting up the most stable and fastest HCL Notes client on...
From A to Z-itrix: Setting up the most stable and fastest HCL Notes client on...
 
Gpdb best practices v a01 20150313
Gpdb best practices v a01 20150313Gpdb best practices v a01 20150313
Gpdb best practices v a01 20150313
 
MongoDB and Azure Databricks
MongoDB and Azure DatabricksMongoDB and Azure Databricks
MongoDB and Azure Databricks
 
Tuning Apache Kafka Connectors for Flink.pptx
Tuning Apache Kafka Connectors for Flink.pptxTuning Apache Kafka Connectors for Flink.pptx
Tuning Apache Kafka Connectors for Flink.pptx
 
Apache Hadoop 3
Apache Hadoop 3Apache Hadoop 3
Apache Hadoop 3
 
Best Practices with PostgreSQL on Solaris
Best Practices with PostgreSQL on SolarisBest Practices with PostgreSQL on Solaris
Best Practices with PostgreSQL on Solaris
 
What Is Hadoop | Hadoop Tutorial For Beginners | Edureka
What Is Hadoop | Hadoop Tutorial For Beginners | EdurekaWhat Is Hadoop | Hadoop Tutorial For Beginners | Edureka
What Is Hadoop | Hadoop Tutorial For Beginners | Edureka
 
Flink Forward Berlin 2018: Stefan Richter - "Tuning Flink for Robustness and ...
Flink Forward Berlin 2018: Stefan Richter - "Tuning Flink for Robustness and ...Flink Forward Berlin 2018: Stefan Richter - "Tuning Flink for Robustness and ...
Flink Forward Berlin 2018: Stefan Richter - "Tuning Flink for Robustness and ...
 
Maximize Greenplum For Any Use Cases Decoupling Compute and Storage - Greenpl...
Maximize Greenplum For Any Use Cases Decoupling Compute and Storage - Greenpl...Maximize Greenplum For Any Use Cases Decoupling Compute and Storage - Greenpl...
Maximize Greenplum For Any Use Cases Decoupling Compute and Storage - Greenpl...
 
Batch Processing at Scale with Flink & Iceberg
Batch Processing at Scale with Flink & IcebergBatch Processing at Scale with Flink & Iceberg
Batch Processing at Scale with Flink & Iceberg
 
The Columnar Era: Leveraging Parquet, Arrow and Kudu for High-Performance Ana...
The Columnar Era: Leveraging Parquet, Arrow and Kudu for High-Performance Ana...The Columnar Era: Leveraging Parquet, Arrow and Kudu for High-Performance Ana...
The Columnar Era: Leveraging Parquet, Arrow and Kudu for High-Performance Ana...
 
Enabling the Active Data Warehouse with Apache Kudu
Enabling the Active Data Warehouse with Apache KuduEnabling the Active Data Warehouse with Apache Kudu
Enabling the Active Data Warehouse with Apache Kudu
 
Big Data and Hadoop
Big Data and HadoopBig Data and Hadoop
Big Data and Hadoop
 
Room 3 - 7 - Nguyễn Như Phúc Huy - Vitastor: a fast and simple Ceph-like bloc...
Room 3 - 7 - Nguyễn Như Phúc Huy - Vitastor: a fast and simple Ceph-like bloc...Room 3 - 7 - Nguyễn Như Phúc Huy - Vitastor: a fast and simple Ceph-like bloc...
Room 3 - 7 - Nguyễn Như Phúc Huy - Vitastor: a fast and simple Ceph-like bloc...
 
Where is my bottleneck? Performance troubleshooting in Flink
Where is my bottleneck? Performance troubleshooting in FlinkWhere is my bottleneck? Performance troubleshooting in Flink
Where is my bottleneck? Performance troubleshooting in Flink
 
Node Labels in YARN
Node Labels in YARNNode Labels in YARN
Node Labels in YARN
 
Introduction to HADOOP.pdf
Introduction to HADOOP.pdfIntroduction to HADOOP.pdf
Introduction to HADOOP.pdf
 
Running Apache NiFi with Apache Spark : Integration Options
Running Apache NiFi with Apache Spark : Integration OptionsRunning Apache NiFi with Apache Spark : Integration Options
Running Apache NiFi with Apache Spark : Integration Options
 
Hive 3 - a new horizon
Hive 3 - a new horizonHive 3 - a new horizon
Hive 3 - a new horizon
 
HBaseConAsia2018 Keynote 2: Recent Development of HBase in Alibaba and Cloud
HBaseConAsia2018 Keynote 2: Recent Development of HBase in Alibaba and CloudHBaseConAsia2018 Keynote 2: Recent Development of HBase in Alibaba and Cloud
HBaseConAsia2018 Keynote 2: Recent Development of HBase in Alibaba and Cloud
 

Andere mochten auch

Hadoop Hardware @Twitter: Size does matter!
Hadoop Hardware @Twitter: Size does matter!Hadoop Hardware @Twitter: Size does matter!
Hadoop Hardware @Twitter: Size does matter!DataWorks Summit
 
How to Reduce your MTTI/MTTR with a Single Click
How to Reduce your MTTI/MTTR with a Single ClickHow to Reduce your MTTI/MTTR with a Single Click
How to Reduce your MTTI/MTTR with a Single ClickSumo Logic
 
White Belt DMAIC Project Line G MTTR
White Belt DMAIC Project Line G  MTTRWhite Belt DMAIC Project Line G  MTTR
White Belt DMAIC Project Line G MTTRIrfan Rasheed Rana
 
Reliability Centered Maintenance Made Simple
Reliability Centered Maintenance Made SimpleReliability Centered Maintenance Made Simple
Reliability Centered Maintenance Made SimpleRicky Smith CMRP, CMRT
 
Reducing MTTR and False Escalations: Event Correlation at LinkedIn
Reducing MTTR and False Escalations: Event Correlation at LinkedInReducing MTTR and False Escalations: Event Correlation at LinkedIn
Reducing MTTR and False Escalations: Event Correlation at LinkedInMichael Kehoe
 
Best Practices in Maintenance and Reliability
Best Practices in Maintenance and ReliabilityBest Practices in Maintenance and Reliability
Best Practices in Maintenance and ReliabilityRicky Smith CMRP, CMRT
 
ELEVATOR MAINTENANCE CONTROL PROGRAM IN WASHINGTON STATE
ELEVATOR MAINTENANCE CONTROL PROGRAM IN WASHINGTON STATEELEVATOR MAINTENANCE CONTROL PROGRAM IN WASHINGTON STATE
ELEVATOR MAINTENANCE CONTROL PROGRAM IN WASHINGTON STATELee Chain
 
8219 worrawan-make a story
8219 worrawan-make a story8219 worrawan-make a story
8219 worrawan-make a storybhawanaskst
 
Breakdowns Happen: Factoring Downtime Into Your Simulation
Breakdowns Happen: Factoring Downtime Into Your SimulationBreakdowns Happen: Factoring Downtime Into Your Simulation
Breakdowns Happen: Factoring Downtime Into Your Simulationbussylee25
 
WHAT'S IN YOUR ELEVATOR MAINTENANCE CONTRACT?
WHAT'S IN YOUR ELEVATOR MAINTENANCE CONTRACT?WHAT'S IN YOUR ELEVATOR MAINTENANCE CONTRACT?
WHAT'S IN YOUR ELEVATOR MAINTENANCE CONTRACT?Lee Chain
 
Maintenance Metrics that Matter
Maintenance Metrics that MatterMaintenance Metrics that Matter
Maintenance Metrics that MattereMaint Enterprises
 
Improving Incident Response: Building a More Efficient IT Infrastructure
Improving Incident Response: Building a More Efficient IT InfrastructureImproving Incident Response: Building a More Efficient IT Infrastructure
Improving Incident Response: Building a More Efficient IT InfrastructureEmulex Corporation
 
Engineering the New LinkedIn Profile
Engineering the New LinkedIn ProfileEngineering the New LinkedIn Profile
Engineering the New LinkedIn ProfileJosh Clemm
 
Intrinsic Methods in HotSpot VM
Intrinsic Methods in HotSpot VMIntrinsic Methods in HotSpot VM
Intrinsic Methods in HotSpot VMKris Mok
 
Getting Hired: How to Get a Job as a Product Manager
Getting Hired: How to Get a Job as a Product ManagerGetting Hired: How to Get a Job as a Product Manager
Getting Hired: How to Get a Job as a Product ManagerJason Shah
 
Unit 9 implementing the reliability strategy
Unit 9  implementing the reliability strategyUnit 9  implementing the reliability strategy
Unit 9 implementing the reliability strategyCharlton Inao
 
Asset Reliability Begins With Your Operators
Asset Reliability Begins With Your OperatorsAsset Reliability Begins With Your Operators
Asset Reliability Begins With Your OperatorsRicky Smith CMRP, CMRT
 

Andere mochten auch (20)

Hadoop Hardware @Twitter: Size does matter!
Hadoop Hardware @Twitter: Size does matter!Hadoop Hardware @Twitter: Size does matter!
Hadoop Hardware @Twitter: Size does matter!
 
How to Reduce your MTTI/MTTR with a Single Click
How to Reduce your MTTI/MTTR with a Single ClickHow to Reduce your MTTI/MTTR with a Single Click
How to Reduce your MTTI/MTTR with a Single Click
 
White Belt DMAIC Project Line G MTTR
White Belt DMAIC Project Line G  MTTRWhite Belt DMAIC Project Line G  MTTR
White Belt DMAIC Project Line G MTTR
 
Reliability Centered Maintenance Made Simple
Reliability Centered Maintenance Made SimpleReliability Centered Maintenance Made Simple
Reliability Centered Maintenance Made Simple
 
Reliability centered maintenance
Reliability centered maintenanceReliability centered maintenance
Reliability centered maintenance
 
Reducing MTTR and False Escalations: Event Correlation at LinkedIn
Reducing MTTR and False Escalations: Event Correlation at LinkedInReducing MTTR and False Escalations: Event Correlation at LinkedIn
Reducing MTTR and False Escalations: Event Correlation at LinkedIn
 
Best Practices in Maintenance and Reliability
Best Practices in Maintenance and ReliabilityBest Practices in Maintenance and Reliability
Best Practices in Maintenance and Reliability
 
ELEVATOR MAINTENANCE CONTROL PROGRAM IN WASHINGTON STATE
ELEVATOR MAINTENANCE CONTROL PROGRAM IN WASHINGTON STATEELEVATOR MAINTENANCE CONTROL PROGRAM IN WASHINGTON STATE
ELEVATOR MAINTENANCE CONTROL PROGRAM IN WASHINGTON STATE
 
8219 worrawan-make a story
8219 worrawan-make a story8219 worrawan-make a story
8219 worrawan-make a story
 
Breakdowns Happen: Factoring Downtime Into Your Simulation
Breakdowns Happen: Factoring Downtime Into Your SimulationBreakdowns Happen: Factoring Downtime Into Your Simulation
Breakdowns Happen: Factoring Downtime Into Your Simulation
 
WHAT'S IN YOUR ELEVATOR MAINTENANCE CONTRACT?
WHAT'S IN YOUR ELEVATOR MAINTENANCE CONTRACT?WHAT'S IN YOUR ELEVATOR MAINTENANCE CONTRACT?
WHAT'S IN YOUR ELEVATOR MAINTENANCE CONTRACT?
 
Maintenance Metrics that Matter
Maintenance Metrics that MatterMaintenance Metrics that Matter
Maintenance Metrics that Matter
 
Improving Incident Response: Building a More Efficient IT Infrastructure
Improving Incident Response: Building a More Efficient IT InfrastructureImproving Incident Response: Building a More Efficient IT Infrastructure
Improving Incident Response: Building a More Efficient IT Infrastructure
 
Engineering the New LinkedIn Profile
Engineering the New LinkedIn ProfileEngineering the New LinkedIn Profile
Engineering the New LinkedIn Profile
 
Intrinsic Methods in HotSpot VM
Intrinsic Methods in HotSpot VMIntrinsic Methods in HotSpot VM
Intrinsic Methods in HotSpot VM
 
Getting Hired: How to Get a Job as a Product Manager
Getting Hired: How to Get a Job as a Product ManagerGetting Hired: How to Get a Job as a Product Manager
Getting Hired: How to Get a Job as a Product Manager
 
How to measure reliability 2
How to measure reliability 2How to measure reliability 2
How to measure reliability 2
 
Unit 9 implementing the reliability strategy
Unit 9  implementing the reliability strategyUnit 9  implementing the reliability strategy
Unit 9 implementing the reliability strategy
 
Asset Reliability Begins With Your Operators
Asset Reliability Begins With Your OperatorsAsset Reliability Begins With Your Operators
Asset Reliability Begins With Your Operators
 
How to measure reliability
How to measure reliabilityHow to measure reliability
How to measure reliability
 

Ähnlich wie HBase: How to get MTTR below 1 minute

HBaseCon 2013: How to Get the MTTR Below 1 Minute and More
HBaseCon 2013: How to Get the MTTR Below 1 Minute and MoreHBaseCon 2013: How to Get the MTTR Below 1 Minute and More
HBaseCon 2013: How to Get the MTTR Below 1 Minute and MoreCloudera, Inc.
 
HBASE by Nicolas Liochon - Meetup HUGFR du 22 Sept 2014
HBASE by  Nicolas Liochon - Meetup HUGFR du 22 Sept 2014HBASE by  Nicolas Liochon - Meetup HUGFR du 22 Sept 2014
HBASE by Nicolas Liochon - Meetup HUGFR du 22 Sept 2014Modern Data Stack France
 
Eventual Consistency @WalmartLabs with Kafka, Avro, SolrCloud and Hadoop
Eventual Consistency @WalmartLabs with Kafka, Avro, SolrCloud and HadoopEventual Consistency @WalmartLabs with Kafka, Avro, SolrCloud and Hadoop
Eventual Consistency @WalmartLabs with Kafka, Avro, SolrCloud and HadoopAyon Sinha
 
HBase Low Latency, StrataNYC 2014
HBase Low Latency, StrataNYC 2014HBase Low Latency, StrataNYC 2014
HBase Low Latency, StrataNYC 2014Nick Dimiduk
 
HBaseCon2017 Improving HBase availability in a multi tenant environment
HBaseCon2017 Improving HBase availability in a multi tenant environmentHBaseCon2017 Improving HBase availability in a multi tenant environment
HBaseCon2017 Improving HBase availability in a multi tenant environmentHBaseCon
 
Apache HBase Low Latency
Apache HBase Low LatencyApache HBase Low Latency
Apache HBase Low LatencyNick Dimiduk
 
DevOps throughout time
DevOps throughout timeDevOps throughout time
DevOps throughout timeHany Fahim
 
Architectural Overview of MapR's Apache Hadoop Distribution
Architectural Overview of MapR's Apache Hadoop DistributionArchitectural Overview of MapR's Apache Hadoop Distribution
Architectural Overview of MapR's Apache Hadoop Distributionmcsrivas
 
Scylla Summit 2016: Outbrain Case Study - Lowering Latency While Doing 20X IO...
Scylla Summit 2016: Outbrain Case Study - Lowering Latency While Doing 20X IO...Scylla Summit 2016: Outbrain Case Study - Lowering Latency While Doing 20X IO...
Scylla Summit 2016: Outbrain Case Study - Lowering Latency While Doing 20X IO...ScyllaDB
 
Apache HBase Performance Tuning
Apache HBase Performance TuningApache HBase Performance Tuning
Apache HBase Performance TuningLars Hofhansl
 
HBaseCon 2015: HBase Performance Tuning @ Salesforce
HBaseCon 2015: HBase Performance Tuning @ SalesforceHBaseCon 2015: HBase Performance Tuning @ Salesforce
HBaseCon 2015: HBase Performance Tuning @ SalesforceHBaseCon
 
On The Building Of A PostgreSQL Cluster
On The Building Of A PostgreSQL ClusterOn The Building Of A PostgreSQL Cluster
On The Building Of A PostgreSQL ClusterSrihari Sriraman
 
Avoiding the ring of death
Avoiding the ring of deathAvoiding the ring of death
Avoiding the ring of deathAishvarya Verma
 
02 2017 emea_roadshow_milan_ha
02 2017 emea_roadshow_milan_ha02 2017 emea_roadshow_milan_ha
02 2017 emea_roadshow_milan_hamlraviol
 
Choosing the right high availability strategy
Choosing the right high availability strategyChoosing the right high availability strategy
Choosing the right high availability strategyMariaDB plc
 
Memcached Presentation
Memcached PresentationMemcached Presentation
Memcached PresentationAsif Ali
 
HBase: Where Online Meets Low Latency
HBase: Where Online Meets Low LatencyHBase: Where Online Meets Low Latency
HBase: Where Online Meets Low LatencyHBaseCon
 
MariaDB Server Performance Tuning & Optimization
MariaDB Server Performance Tuning & OptimizationMariaDB Server Performance Tuning & Optimization
MariaDB Server Performance Tuning & OptimizationMariaDB plc
 
Choosing the right high availability strategy
Choosing the right high availability strategyChoosing the right high availability strategy
Choosing the right high availability strategyMariaDB plc
 

Ähnlich wie HBase: How to get MTTR below 1 minute (20)

HBaseCon 2013: How to Get the MTTR Below 1 Minute and More
HBaseCon 2013: How to Get the MTTR Below 1 Minute and MoreHBaseCon 2013: How to Get the MTTR Below 1 Minute and More
HBaseCon 2013: How to Get the MTTR Below 1 Minute and More
 
HBASE by Nicolas Liochon - Meetup HUGFR du 22 Sept 2014
HBASE by  Nicolas Liochon - Meetup HUGFR du 22 Sept 2014HBASE by  Nicolas Liochon - Meetup HUGFR du 22 Sept 2014
HBASE by Nicolas Liochon - Meetup HUGFR du 22 Sept 2014
 
Eventual Consistency @WalmartLabs with Kafka, Avro, SolrCloud and Hadoop
Eventual Consistency @WalmartLabs with Kafka, Avro, SolrCloud and HadoopEventual Consistency @WalmartLabs with Kafka, Avro, SolrCloud and Hadoop
Eventual Consistency @WalmartLabs with Kafka, Avro, SolrCloud and Hadoop
 
HBase Low Latency, StrataNYC 2014
HBase Low Latency, StrataNYC 2014HBase Low Latency, StrataNYC 2014
HBase Low Latency, StrataNYC 2014
 
HBaseCon2017 Improving HBase availability in a multi tenant environment
HBaseCon2017 Improving HBase availability in a multi tenant environmentHBaseCon2017 Improving HBase availability in a multi tenant environment
HBaseCon2017 Improving HBase availability in a multi tenant environment
 
Apache HBase Low Latency
Apache HBase Low LatencyApache HBase Low Latency
Apache HBase Low Latency
 
DevOps throughout time
DevOps throughout timeDevOps throughout time
DevOps throughout time
 
Architectural Overview of MapR's Apache Hadoop Distribution
Architectural Overview of MapR's Apache Hadoop DistributionArchitectural Overview of MapR's Apache Hadoop Distribution
Architectural Overview of MapR's Apache Hadoop Distribution
 
Storm 2012 03-29
Storm 2012 03-29Storm 2012 03-29
Storm 2012 03-29
 
Scylla Summit 2016: Outbrain Case Study - Lowering Latency While Doing 20X IO...
Scylla Summit 2016: Outbrain Case Study - Lowering Latency While Doing 20X IO...Scylla Summit 2016: Outbrain Case Study - Lowering Latency While Doing 20X IO...
Scylla Summit 2016: Outbrain Case Study - Lowering Latency While Doing 20X IO...
 
Apache HBase Performance Tuning
Apache HBase Performance TuningApache HBase Performance Tuning
Apache HBase Performance Tuning
 
HBaseCon 2015: HBase Performance Tuning @ Salesforce
HBaseCon 2015: HBase Performance Tuning @ SalesforceHBaseCon 2015: HBase Performance Tuning @ Salesforce
HBaseCon 2015: HBase Performance Tuning @ Salesforce
 
On The Building Of A PostgreSQL Cluster
On The Building Of A PostgreSQL ClusterOn The Building Of A PostgreSQL Cluster
On The Building Of A PostgreSQL Cluster
 
Avoiding the ring of death
Avoiding the ring of deathAvoiding the ring of death
Avoiding the ring of death
 
02 2017 emea_roadshow_milan_ha
02 2017 emea_roadshow_milan_ha02 2017 emea_roadshow_milan_ha
02 2017 emea_roadshow_milan_ha
 
Choosing the right high availability strategy
Choosing the right high availability strategyChoosing the right high availability strategy
Choosing the right high availability strategy
 
Memcached Presentation
Memcached PresentationMemcached Presentation
Memcached Presentation
 
HBase: Where Online Meets Low Latency
HBase: Where Online Meets Low LatencyHBase: Where Online Meets Low Latency
HBase: Where Online Meets Low Latency
 
MariaDB Server Performance Tuning & Optimization
MariaDB Server Performance Tuning & OptimizationMariaDB Server Performance Tuning & Optimization
MariaDB Server Performance Tuning & Optimization
 
Choosing the right high availability strategy
Choosing the right high availability strategyChoosing the right high availability strategy
Choosing the right high availability strategy
 

Mehr von Hortonworks

Hortonworks DataFlow (HDF) 3.3 - Taking Stream Processing to the Next Level
Hortonworks DataFlow (HDF) 3.3 - Taking Stream Processing to the Next LevelHortonworks DataFlow (HDF) 3.3 - Taking Stream Processing to the Next Level
Hortonworks DataFlow (HDF) 3.3 - Taking Stream Processing to the Next LevelHortonworks
 
IoT Predictions for 2019 and Beyond: Data at the Heart of Your IoT Strategy
IoT Predictions for 2019 and Beyond: Data at the Heart of Your IoT StrategyIoT Predictions for 2019 and Beyond: Data at the Heart of Your IoT Strategy
IoT Predictions for 2019 and Beyond: Data at the Heart of Your IoT StrategyHortonworks
 
Getting the Most Out of Your Data in the Cloud with Cloudbreak
Getting the Most Out of Your Data in the Cloud with CloudbreakGetting the Most Out of Your Data in the Cloud with Cloudbreak
Getting the Most Out of Your Data in the Cloud with CloudbreakHortonworks
 
Johns Hopkins - Using Hadoop to Secure Access Log Events
Johns Hopkins - Using Hadoop to Secure Access Log EventsJohns Hopkins - Using Hadoop to Secure Access Log Events
Johns Hopkins - Using Hadoop to Secure Access Log EventsHortonworks
 
Catch a Hacker in Real-Time: Live Visuals of Bots and Bad Guys
Catch a Hacker in Real-Time: Live Visuals of Bots and Bad GuysCatch a Hacker in Real-Time: Live Visuals of Bots and Bad Guys
Catch a Hacker in Real-Time: Live Visuals of Bots and Bad GuysHortonworks
 
HDF 3.2 - What's New
HDF 3.2 - What's NewHDF 3.2 - What's New
HDF 3.2 - What's NewHortonworks
 
Curing Kafka Blindness with Hortonworks Streams Messaging Manager
Curing Kafka Blindness with Hortonworks Streams Messaging ManagerCuring Kafka Blindness with Hortonworks Streams Messaging Manager
Curing Kafka Blindness with Hortonworks Streams Messaging ManagerHortonworks
 
Interpretation Tool for Genomic Sequencing Data in Clinical Environments
Interpretation Tool for Genomic Sequencing Data in Clinical EnvironmentsInterpretation Tool for Genomic Sequencing Data in Clinical Environments
Interpretation Tool for Genomic Sequencing Data in Clinical EnvironmentsHortonworks
 
IBM+Hortonworks = Transformation of the Big Data Landscape
IBM+Hortonworks = Transformation of the Big Data LandscapeIBM+Hortonworks = Transformation of the Big Data Landscape
IBM+Hortonworks = Transformation of the Big Data LandscapeHortonworks
 
Premier Inside-Out: Apache Druid
Premier Inside-Out: Apache DruidPremier Inside-Out: Apache Druid
Premier Inside-Out: Apache DruidHortonworks
 
Accelerating Data Science and Real Time Analytics at Scale
Accelerating Data Science and Real Time Analytics at ScaleAccelerating Data Science and Real Time Analytics at Scale
Accelerating Data Science and Real Time Analytics at ScaleHortonworks
 
TIME SERIES: APPLYING ADVANCED ANALYTICS TO INDUSTRIAL PROCESS DATA
TIME SERIES: APPLYING ADVANCED ANALYTICS TO INDUSTRIAL PROCESS DATATIME SERIES: APPLYING ADVANCED ANALYTICS TO INDUSTRIAL PROCESS DATA
TIME SERIES: APPLYING ADVANCED ANALYTICS TO INDUSTRIAL PROCESS DATAHortonworks
 
Blockchain with Machine Learning Powered by Big Data: Trimble Transportation ...
Blockchain with Machine Learning Powered by Big Data: Trimble Transportation ...Blockchain with Machine Learning Powered by Big Data: Trimble Transportation ...
Blockchain with Machine Learning Powered by Big Data: Trimble Transportation ...Hortonworks
 
Delivering Real-Time Streaming Data for Healthcare Customers: Clearsense
Delivering Real-Time Streaming Data for Healthcare Customers: ClearsenseDelivering Real-Time Streaming Data for Healthcare Customers: Clearsense
Delivering Real-Time Streaming Data for Healthcare Customers: ClearsenseHortonworks
 
Making Enterprise Big Data Small with Ease
Making Enterprise Big Data Small with EaseMaking Enterprise Big Data Small with Ease
Making Enterprise Big Data Small with EaseHortonworks
 
Webinewbie to Webinerd in 30 Days - Webinar World Presentation
Webinewbie to Webinerd in 30 Days - Webinar World PresentationWebinewbie to Webinerd in 30 Days - Webinar World Presentation
Webinewbie to Webinerd in 30 Days - Webinar World PresentationHortonworks
 
Driving Digital Transformation Through Global Data Management
Driving Digital Transformation Through Global Data ManagementDriving Digital Transformation Through Global Data Management
Driving Digital Transformation Through Global Data ManagementHortonworks
 
HDF 3.1 pt. 2: A Technical Deep-Dive on New Streaming Features
HDF 3.1 pt. 2: A Technical Deep-Dive on New Streaming FeaturesHDF 3.1 pt. 2: A Technical Deep-Dive on New Streaming Features
HDF 3.1 pt. 2: A Technical Deep-Dive on New Streaming FeaturesHortonworks
 
Hortonworks DataFlow (HDF) 3.1 - Redefining Data-In-Motion with Modern Data A...
Hortonworks DataFlow (HDF) 3.1 - Redefining Data-In-Motion with Modern Data A...Hortonworks DataFlow (HDF) 3.1 - Redefining Data-In-Motion with Modern Data A...
Hortonworks DataFlow (HDF) 3.1 - Redefining Data-In-Motion with Modern Data A...Hortonworks
 
Unlock Value from Big Data with Apache NiFi and Streaming CDC
Unlock Value from Big Data with Apache NiFi and Streaming CDCUnlock Value from Big Data with Apache NiFi and Streaming CDC
Unlock Value from Big Data with Apache NiFi and Streaming CDCHortonworks
 

Mehr von Hortonworks (20)

Hortonworks DataFlow (HDF) 3.3 - Taking Stream Processing to the Next Level
Hortonworks DataFlow (HDF) 3.3 - Taking Stream Processing to the Next LevelHortonworks DataFlow (HDF) 3.3 - Taking Stream Processing to the Next Level
Hortonworks DataFlow (HDF) 3.3 - Taking Stream Processing to the Next Level
 
IoT Predictions for 2019 and Beyond: Data at the Heart of Your IoT Strategy
IoT Predictions for 2019 and Beyond: Data at the Heart of Your IoT StrategyIoT Predictions for 2019 and Beyond: Data at the Heart of Your IoT Strategy
IoT Predictions for 2019 and Beyond: Data at the Heart of Your IoT Strategy
 
Getting the Most Out of Your Data in the Cloud with Cloudbreak
Getting the Most Out of Your Data in the Cloud with CloudbreakGetting the Most Out of Your Data in the Cloud with Cloudbreak
Getting the Most Out of Your Data in the Cloud with Cloudbreak
 
Johns Hopkins - Using Hadoop to Secure Access Log Events
Johns Hopkins - Using Hadoop to Secure Access Log EventsJohns Hopkins - Using Hadoop to Secure Access Log Events
Johns Hopkins - Using Hadoop to Secure Access Log Events
 
Catch a Hacker in Real-Time: Live Visuals of Bots and Bad Guys
Catch a Hacker in Real-Time: Live Visuals of Bots and Bad GuysCatch a Hacker in Real-Time: Live Visuals of Bots and Bad Guys
Catch a Hacker in Real-Time: Live Visuals of Bots and Bad Guys
 
HDF 3.2 - What's New
HDF 3.2 - What's NewHDF 3.2 - What's New
HDF 3.2 - What's New
 
Curing Kafka Blindness with Hortonworks Streams Messaging Manager
Curing Kafka Blindness with Hortonworks Streams Messaging ManagerCuring Kafka Blindness with Hortonworks Streams Messaging Manager
Curing Kafka Blindness with Hortonworks Streams Messaging Manager
 
Interpretation Tool for Genomic Sequencing Data in Clinical Environments
Interpretation Tool for Genomic Sequencing Data in Clinical EnvironmentsInterpretation Tool for Genomic Sequencing Data in Clinical Environments
Interpretation Tool for Genomic Sequencing Data in Clinical Environments
 
IBM+Hortonworks = Transformation of the Big Data Landscape
IBM+Hortonworks = Transformation of the Big Data LandscapeIBM+Hortonworks = Transformation of the Big Data Landscape
IBM+Hortonworks = Transformation of the Big Data Landscape
 
Premier Inside-Out: Apache Druid
Premier Inside-Out: Apache DruidPremier Inside-Out: Apache Druid
Premier Inside-Out: Apache Druid
 
Accelerating Data Science and Real Time Analytics at Scale
Accelerating Data Science and Real Time Analytics at ScaleAccelerating Data Science and Real Time Analytics at Scale
Accelerating Data Science and Real Time Analytics at Scale
 
TIME SERIES: APPLYING ADVANCED ANALYTICS TO INDUSTRIAL PROCESS DATA
TIME SERIES: APPLYING ADVANCED ANALYTICS TO INDUSTRIAL PROCESS DATATIME SERIES: APPLYING ADVANCED ANALYTICS TO INDUSTRIAL PROCESS DATA
TIME SERIES: APPLYING ADVANCED ANALYTICS TO INDUSTRIAL PROCESS DATA
 
Blockchain with Machine Learning Powered by Big Data: Trimble Transportation ...
Blockchain with Machine Learning Powered by Big Data: Trimble Transportation ...Blockchain with Machine Learning Powered by Big Data: Trimble Transportation ...
Blockchain with Machine Learning Powered by Big Data: Trimble Transportation ...
 
Delivering Real-Time Streaming Data for Healthcare Customers: Clearsense
Delivering Real-Time Streaming Data for Healthcare Customers: ClearsenseDelivering Real-Time Streaming Data for Healthcare Customers: Clearsense
Delivering Real-Time Streaming Data for Healthcare Customers: Clearsense
 
Making Enterprise Big Data Small with Ease
Making Enterprise Big Data Small with EaseMaking Enterprise Big Data Small with Ease
Making Enterprise Big Data Small with Ease
 
Webinewbie to Webinerd in 30 Days - Webinar World Presentation
Webinewbie to Webinerd in 30 Days - Webinar World PresentationWebinewbie to Webinerd in 30 Days - Webinar World Presentation
Webinewbie to Webinerd in 30 Days - Webinar World Presentation
 
Driving Digital Transformation Through Global Data Management
Driving Digital Transformation Through Global Data ManagementDriving Digital Transformation Through Global Data Management
Driving Digital Transformation Through Global Data Management
 
HDF 3.1 pt. 2: A Technical Deep-Dive on New Streaming Features
HDF 3.1 pt. 2: A Technical Deep-Dive on New Streaming FeaturesHDF 3.1 pt. 2: A Technical Deep-Dive on New Streaming Features
HDF 3.1 pt. 2: A Technical Deep-Dive on New Streaming Features
 
Hortonworks DataFlow (HDF) 3.1 - Redefining Data-In-Motion with Modern Data A...
Hortonworks DataFlow (HDF) 3.1 - Redefining Data-In-Motion with Modern Data A...Hortonworks DataFlow (HDF) 3.1 - Redefining Data-In-Motion with Modern Data A...
Hortonworks DataFlow (HDF) 3.1 - Redefining Data-In-Motion with Modern Data A...
 
Unlock Value from Big Data with Apache NiFi and Streaming CDC
Unlock Value from Big Data with Apache NiFi and Streaming CDCUnlock Value from Big Data with Apache NiFi and Streaming CDC
Unlock Value from Big Data with Apache NiFi and Streaming CDC
 

Kürzlich hochgeladen

The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxLoriGlavin3
 
Assure Ecommerce and Retail Operations Uptime with ThousandEyes
Assure Ecommerce and Retail Operations Uptime with ThousandEyesAssure Ecommerce and Retail Operations Uptime with ThousandEyes
Assure Ecommerce and Retail Operations Uptime with ThousandEyesThousandEyes
 
Manual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance AuditManual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance AuditSkynet Technologies
 
Sample pptx for embedding into website for demo
Sample pptx for embedding into website for demoSample pptx for embedding into website for demo
Sample pptx for embedding into website for demoHarshalMandlekar2
 
Generative AI for Technical Writer or Information Developers
Generative AI for Technical Writer or Information DevelopersGenerative AI for Technical Writer or Information Developers
Generative AI for Technical Writer or Information DevelopersRaghuram Pandurangan
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersNicole Novielli
 
How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.Curtis Poe
 
UiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to HeroUiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to HeroUiPathCommunity
 
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Alkin Tezuysal
 
How to write a Business Continuity Plan
How to write a Business Continuity PlanHow to write a Business Continuity Plan
How to write a Business Continuity PlanDatabarracks
 
Scale your database traffic with Read & Write split using MySQL Router
Scale your database traffic with Read & Write split using MySQL RouterScale your database traffic with Read & Write split using MySQL Router
Scale your database traffic with Read & Write split using MySQL RouterMydbops
 
From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .Alan Dix
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxLoriGlavin3
 
The State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptxThe State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptxLoriGlavin3
 
Digital Identity is Under Attack: FIDO Paris Seminar.pptx
Digital Identity is Under Attack: FIDO Paris Seminar.pptxDigital Identity is Under Attack: FIDO Paris Seminar.pptx
Digital Identity is Under Attack: FIDO Paris Seminar.pptxLoriGlavin3
 
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality AssuranceInflectra
 
Generative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdfGenerative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdfIngrid Airi González
 
Connecting the Dots for Information Discovery.pdf
Connecting the Dots for Information Discovery.pdfConnecting the Dots for Information Discovery.pdf
Connecting the Dots for Information Discovery.pdfNeo4j
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch TuesdayIvanti
 
Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...Farhan Tariq
 

Kürzlich hochgeladen (20)

The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
 
Assure Ecommerce and Retail Operations Uptime with ThousandEyes
Assure Ecommerce and Retail Operations Uptime with ThousandEyesAssure Ecommerce and Retail Operations Uptime with ThousandEyes
Assure Ecommerce and Retail Operations Uptime with ThousandEyes
 
Manual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance AuditManual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance Audit
 
Sample pptx for embedding into website for demo
Sample pptx for embedding into website for demoSample pptx for embedding into website for demo
Sample pptx for embedding into website for demo
 
Generative AI for Technical Writer or Information Developers
Generative AI for Technical Writer or Information DevelopersGenerative AI for Technical Writer or Information Developers
Generative AI for Technical Writer or Information Developers
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software Developers
 
How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.
 
UiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to HeroUiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to Hero
 
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
 
How to write a Business Continuity Plan
How to write a Business Continuity PlanHow to write a Business Continuity Plan
How to write a Business Continuity Plan
 
Scale your database traffic with Read & Write split using MySQL Router
Scale your database traffic with Read & Write split using MySQL RouterScale your database traffic with Read & Write split using MySQL Router
Scale your database traffic with Read & Write split using MySQL Router
 
From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
 
The State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptxThe State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptx
 
Digital Identity is Under Attack: FIDO Paris Seminar.pptx
Digital Identity is Under Attack: FIDO Paris Seminar.pptxDigital Identity is Under Attack: FIDO Paris Seminar.pptx
Digital Identity is Under Attack: FIDO Paris Seminar.pptx
 
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
 
Generative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdfGenerative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdf
 
Connecting the Dots for Information Discovery.pdf
Connecting the Dots for Information Discovery.pdfConnecting the Dots for Information Discovery.pdf
Connecting the Dots for Information Discovery.pdf
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch Tuesday
 
Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...
 

HBase: How to get MTTR below 1 minute

  • 1. How  to  get  the  MTTR  below  1   minute  and  more   Devaraj  Das   (ddas@hortonworks.com)   Nicolas  Liochon   (nkeywal@gmail.com)  
  • 2. Outline   •  What  is  this?  Why  are  we  talking  about  this   topic?  Why  it  ma>ers?  ….   •  HBase  Recovery  –  an  overview   •  HDFS  issues   •  Beyond  MTTR  (Performance  post  recovery)   •  Conclusion  /  Future  /  Q  &  A  
  • 3. What  is  MTTR?  Why  its  important?  …   •  Mean  Time  To  Recovery  -­‐>  Average  Pme  required   to  repair  a  failed  component  (Courtesy:  Wikipedia)   •  Enterprises  want  an  MTTR  of  ZERO   –  Data  should  always  be  available  with  no  degradaPon   of  perceived  SLAs   –  PracPcally  hard  to  obtain  but  yeah  it’s  a  goal   •  Close  to  Zero-­‐MTTR  is  especially  important  for   HBase   –  Given  it  is  used  in  near  realPme  systems   •  MTTR  in  other  NoSQL  systems  &  Databases  
  • 4. HBase  Basics   •  Strongly  consistent   –  Write  ordered  with  reads   –  Once  wri>en,  the  data  will  stay   •  Built  on  top  of  HDFS   •  When  a  machine  fails  the  cluster  remains   available,  and  its  data  as  well   •  We’re  just  speaking  about  the  piece  of  data  that   was  handled  by  this  machine  
  • 5. Write  path   WAL  –  Write   Ahead  Log   A  write  is   finished  once   wri>en  on  all   HDFS  nodes   The  client   communicated   with  the  region   servers  
  • 6. We’re  in  a  distributed  system   •  You  can’t  disPnguish  a   slow  server  from  a   dead  server   •  Everything,  or,  nearly   everything,  is  based   on  Pmeout   •  Smaller  Pmeouts  means  more  false  posiPve   •  HBase  works  well  with  false  posiPve,  but  they   always  have  a  cost.   •  The  less  the  Pmeouts  the  be>er  
  • 7. HBase  components  for  recovery  
  • 9. Recovery  process   •  Failure  detecPon:  ZooKeeper   heartbeats  the  servers.  Expire   the  session  when  it  does  not   reply   •  Region  assignment:  the  master   reallocates  the  regions  to  the   other  servers   •  Failure  recovery:  read  the  WAL   and  rewrite  the  data  again   •  The  clients  stops  the   connecPon  to  the  dead  server   and  goes  to  the  new  one.   ZK   Heartbeat   Client   Region  Servers,   DataNode   Data  recovery   Master,  RS,  ZK   Region  Assignment  
  • 10. So….   •  Detect  the  failure  as  fast  as  possible   •  Reassign  as  fast  as  possible   •  Read  /  rewrite  the  WAL  as  fast  as  possible   •  That’s  obvious  
  • 11. The  obvious  –  failure  detecPon   •  Failure  detecPon   –  Set  a  ZooKeeper  Pmeout  to  30s  instead  of  the  old  180s   default.     –  Beware  of  the  GC,  but  lower  values  are  possible.   –  ZooKeeper  detects  the  errors  sooner  than  the  configured   Pmeout   •  0.96     –  HBase  scripts  clean  the  ZK  node  when  the  server  is  kill   -­‐9ed   •  =>  DetecPon  Pme  becomes  0   –  Can  be  used  by  any  monitoring  tool  
  • 12. The  obvious  –  faster  data  recovery   •  Not  so  obvious  actually   •  Already  distributed  since  0.92   –  The  large  the  cluster  the  be>er.   •  Completely  rewri>en  in  0.96   –  Recovery  itself  rewri>en  in  0.96   –  Will  be  covered  in  the  second  part  
  • 13. The  obvious  –  Faster  assignment   •  Faster  assignment   –  Just  improving  performances   •  Parallelism   •  Speed   –  Globally  ‘much’  faster   –  Backported  to  0.94   •  SPll  possible  to  do  be>er  for  huge  number  of   regions.     •  A  few  seconds  for  most  cases  
  • 14. With  this   •  DetecPon:  from  180s  to  30s   •  Data  recovery:  around  10s   •  Reassignment  :  from  10s  of  seconds  to   seconds  
  • 15. Do  you  think  we’re  be>er  with  this   •  Answer  is  NO   •  Actually,  yes  but  if  and  only  if  HDFS  is  fine   – But  when  you  lose  a  regionserver,  you’ve  just  lost   a  datanode  
  • 16. DataNode  crash  is  expensive!   •  One  replica  of  WAL  edits  is  on  the  crashed  DN   – 33%  of  the  reads  during  the  regionserver  recovery   will  go  to  it   •  Many  writes  will  go  to  it  as  well  (the  smaller   the  cluster,  the  higher  that  probability)   •  NameNode  re-­‐replicates  the  data  (maybe  TBs)   that  was  on  this  node  to  restore  replica  count   – NameNode  does  this  work  only  amer  a  good   Pmeout  (10  minutes  by  default)  
  • 17. HDFS  –  Stale  mode   Live   Stale   Dead   As  today:  used  for  reads  &   writes,  using  locality   Not  used  for  writes,  used  as   last  resort  for  reads   As  today:  not  used.   And  actually,  it’s  be>er  to  do  the  HBase   recovery  before  HDFS  replicates  the  TBs   of  data  of  this  node   30  seconds,  can  be  less.   10  minutes,  don’t  change  this  
  • 18. Results   •  Do  more  read/writes  to  HDFS  during  the   recovery   •  MulPple  failures  are  sPll  possible   – Stale  mode  will  sPll  play  its  role   – And  set  dfs.Pmeout  to  30s   – This  limits  the  effect  of  two  failure  in  a  row.  The   cost  of  the  second  failure  is  30s  if  you  were   unlucky  
  • 19. Are  we  done?   •  We’re  not  bad   •  But  there  is  sPll  something  
  • 20. The  client   You  lem  it  waiPng  on  the  dead  server          
  • 22. The  client   •  You  want  the  client  to  be  paPent   •  Retries  when  the  system  is  already  loaded  is   not  good.     •  You  want  the  client  to  learn  about  region   servers  dying,  and  to  be  able  to  react   immediately.   •  You  want  this  to  scale.  
  • 23. SoluPon   •  The  master  noPfies  the  client   –  A  cheap  mulPcast  message  with  the  “dead  servers”   list.  Sent  5  Pmes  for  safety.   –  Off  by  default.   –  On  recepPon,  the  client  stops  immediately  waiPng  on   the  TCP  connecPon.  You  can  now  enjoy  large   hbase.rpc.Pmeout  
  • 24. Full  workflow   t0   t1   t2   t3   Client  reads   and  writes   RegionServer  serving   reads  and  writes   RegionServer  crashes   Affected  regions   reassigned   Client  writes   Data  recovered   Client  reads   and  writes  t4  
  • 25. Are  we  done   •  In  a  way,  yes   – There  is  a  lot  of  things  around  asynchronous   writes,  reads  during  recovery   – Will  be  for  another  Pme,  but  there  will  be  some   nice  things  in  0.96   •  And  a  couple  of  them  is  presented  in  the   second  part  of  this  talk!  
  • 26. Faster  recovery   •  Previous  algo   –  Read  the  WAL  files   –  Write  new  Hfiles   –  Tell  the  region  server  it  got  new  Hfiles   •  Put  pressure  on  namenode   –  Remember:  don’t  put  pressure  on  the  namenode   •  New  algo:   –  Read  the  WAL   –  Write  to  the  regionserver   –  We’re  done  (have  seen  great  improvements  in  our  tests)   –  TBD:  Assign  the  WAL  to  a  RegionServer  local  to  a  replica  
  • 27. RegionServer0   RegionServer_x   RegionServer_y   WAL-­‐file3   <region2:edit1><region1:edit2>   ……   <region3:edit1>   ……..   WAL-­‐file2   <region2:edit1><region1:edit2>   ……   <region3:edit1>   ……..   WAL-­‐file1   <region2:edit1><region1:edit2>   ……   <region3:edit1>   ……..   HDFS   Splitlog-­‐file-­‐for-­‐region3   <region3:edit1><region1:edit2>   ……   <region3:edit1>   ……..   Splitlog-­‐file-­‐for-­‐region2   <region2:edit1><region1:edit2>   ……   <region2:edit1>   ……..   Splitlog-­‐file-­‐for-­‐region1   <region1:edit1><region1:edit2>   ……   <region1:edit1>   ……..   HDFS   RegionServer3   RegionServer2   RegionServer1   writes   writes   reads   reads   Distributed  log   Split  
  • 28. RegionServer0   RegionServer_x   RegionServer_y   WAL-­‐file3   <region2:edit1><region1:edit2>   ……   <region3:edit1>   ……..   WAL-­‐file2   <region2:edit1><region1:edit2>   ……   <region3:edit1>   ……..   WAL-­‐file1   <region2:edit1><region1:edit2>   ……   <region3:edit1>   ……..   HDFS   Recovered-­‐file-­‐for-­‐region3   <region3:edit1><region1:edit2>   ……   <region3:edit1>   ……..   Recovered-­‐file-­‐for-­‐region2   <region2:edit1><region1:edit2>   ……   <region2:edit1>   ……..   Recovered-­‐file-­‐for-­‐region1   <region1:edit1><region1:edit2>   ……   <region1:edit1>   ……..   HDFS   RegionServer3   RegionServer2   RegionServer1   writes   writes   reads   reads   Distributed  log   Replay   replays  
  • 29. Write  during  recovery   •  Hey,  you  can  write  during  the  WAL  replay   •  Events  stream:  your  new  recovery  Pme  is  the   failure  detecPon  Pme:  max  30s,  likely  less!  
  • 30. MemStore  flush   •  Real  life:  some  tables  are  updated  at  a  given   moment  then  lem  alone   – With  a  non  empty  memstore   – More  data  to  recover   •  It’s  now  possible  to  guarantee  that  we  don’t   have  MemStore  with  old  data   •  Improves  real  life  MTTR   •  Helps  snapshots  
  • 31. .META.   •  .META.   –  There  is  no  –ROOT-­‐  in  0.95/0.96   –  But  .META.  failures  are  criPcal   •  A  lot  of  small  improvements   –  Server  now  says  to  the  client  when  a  region  has   moved  (client  can  avoid  going  to  meta)   •  And  a  big  one   –  .META.  WAL  is  managed  separately  to  allow  an   immediate  recovery  of  META   –  With  the  new  MemStore  flush,  ensure  a  quick   recovery  
  • 32. Data  locality  post  recovery   •  HBase  performance  depends  on  data-­‐locality   •  Amer  a  recovery,  you’ve  lost  it   –  Bad  for  performance   •  Here  comes  region  groups   •  Assign  3  favored  RegionServers  for  every  region   •  On  failures  assign  the  region  to  one  of  the   secondaries   •  The  data-­‐locality  issue  is  minimized  on  failures  
  • 33. Block1   Block2   Block3   Block1   Block2   Rack1   Block3   Block3   Rack2   Rack3   Block1   Block2   Datanode   RegionServer1   Datanode1   RegionServer1   Datanode   RegionServer2   Datanode1   RegionServer1   Datanode   RegionServer3   Block1   Block2   Rack1   Block3   Block3   Rack2   Rack3   Block1   Block2   RegionServer4   Datanode1   RegionServer1   Datanode   RegionServer2   Datanode1   RegionServer1   Datanode   RegionServer3   Reads  Blk1  and   Blk2  remotely   Reads  Blk3   remotely   RegionServer1  serves  three  regions,  and  their  StoreFile  blks  are  sca>ered   across  the  cluster  with  one  replica  local  to  RegionServer1.  
  • 34. Block1   Block2   Block3   Block1   Block2   Rack1   Block3   Block3   Rack2   Rack3   Block1   Block2   Datanode   RegionServer1   Datanode1   RegionServer1   Datanode   RegionServer2   Datanode1   RegionServer1   Datanode   RegionServer3   RegionServer1  serves  three  regions,  and  their  StoreFile  blks  are  placed  on   specific  machines  on  the  other  racks   Block1   Block2   Rack1   Block3   Block3   Rack2   Rack3   Block1   Block2   RegionServer4   Datanode1   RegionServer1   Datanode   RegionServer2   Datanode1   RegionServer1   Datanode   RegionServer3   No  remote  reads   Datanode  
  • 35. Conclusion   •  The  target  was  “from  omen  10  minutes  to   always  less  than  1  minute”   – We’re  almost  there   •  Most  of  it  is  available  in  0.96,  some  parts  were   backported   •  Real  life  tesPng  of  the  improvements  in   progress   •  Room  for  more  improvements  
  • 36. Q  &  A   Thanks!