SlideShare ist ein Scribd-Unternehmen logo
1 von 26
Future Information Growth
           AND
 Storage Device Reliability



        Andrei Khurshudov
         Seagate Technology
                2007
The History of Data Storage




• Storage media: charcoal and dirt on stone
• Data type: analog (image)
• Storage life: >17,000 years (in a sealed dry                          ‘Diamond Sutra’ (the world’s earliest
                                                                        complete survival of a dated printed book),
cave)
                                                                        AD 868

                                                                        Storage media: ink on paper

                                                                        Data type: analog (images, characters)

                                                                        Storage life: >1,100 years (sealed in a cave)

                                              Andrei Khurshudov, 2007
The History of Computer Data Storage                                                                        1.8” Perpendicular
                                                                                                                   2005
                                                                  5.25” drive    2.5” drive   1.8” drive
                RAMAC Hard disk drive
                                               3340 Winchester,      1980                        1991
                                                                                    1988
                       1956
                                                    1962
                                                                                                                        Hybrid




                                                                                                           Jazz
                                                                                                   Zip
                                                 The floppy              3.5” drive                                        SSD
                               Magnetic drum                                1983


                                                                                                                  Blue-ray/HD DVD

               Don’t know how to sell more
               storage…


                                                                                                   DVD
                                                                  CD ROM
       Direct access to data              Magnetic Tape                                             Holographic Disk disk
                                                                   1980
                                                                                CD/DVD                       Holographic

                                         Sequential access to data

                                                                                                              Need more storage!
                 Punch cards

                                                     Compact Cassette


                               Magnetic tape
              Punched tape




1940          1950               1960           1970              1980                1990                 2000                  2010   2020


 quot;Do not fold, spindle or mutilate”


                                                           Andrei Khurshudov, 2007
The First HDD is Born




                                                            • Stands for quot;Random Access Method of
                                                            Accounting and Controlquot;
                                                            • Born: 1956
                                                            • Capacity: 5 MB
                                                            • Disk diameter: 24”
                                                            • Recording surfaces: 100
                                                            • Tracks/surface: 100
                                                            • RPM: 1200
                                                            • Weight: >1 ton
                                                            • Cost: leased for $3,200 per month



“While the storage capacity of the drive could have been increased above five megabytes, the
marketing department at IBM was against a larger capacity drive because they didn't know how
to sell a product with more storage (source: Currie Munce, VP, IBM Research)
                                  Andrei Khurshudov, 2007
Modern Disk Drive

About 50 years old                                      Runs faster with every year…
Mass-produced electro-mechanical device                 2006 total industry output >400M drives
Utilizes principles of magnetic recording               Most recent products utilize PMR
Relies on a flying magnetic element                     Typical mechanical separation ~5-10 nm
Available in several standard form factors              1”, 1.8”, 2.5”, 3.5”
Designed for several distinct markets                   Desktop, Enterprise, Mobile, HH, CE
Uses various computer interfaces                        PATA, SATA, SAS, SCSI, FCAL
Historically high data density growth rate              CAGR of 30% to 50% over the last decades
Experiences constant cost pressure                      Cost of GB is under $0.5 and falling
Always under attack from disruptive                     Destroys or assimilates competition for 50
technologies                                            years
Continually expands into new markets                    Most recent: CE, automotive, archival
Highly competitive industry                             Darwinian principles in accelerated action*
Industry share leader: Seagate                          ~40% of the total market share



                                                                  * “The Innovator’s Dilemma” by Clayton M. Christensen
                                                                         Innovator’ Dilemma”

                                   Andrei Khurshudov, 2007
Disk Drive Industry Trends
                                                                                                                                                   0.85” drive


                                                                             Source: PC World, The Hard Drive Turns 50



                                           Source: Coughlin Associates




                                                                                                                                    Bear Stearns Technology Conference, 2006




Bear Stearns Technology Conference, 2006




                                                                                                               Ed Grochowski, IBM
                                                                                                                                                                               Ed Grochowski, IBM




                                                                         Drives get denser, smaller, faster, and cheaper
                                                                         Reliability becomes increasingly difficult
                                                                           Andrei Khurshudov, 2007
Yesterday, Today, and Tomorrow
                                               Tomorrow
         Yesterday




           Today




     There’s plenty of room at the bottom!
                     Andrei Khurshudov, 2007
Estimated Number of Units Shipped

                                  900,000
                                  800,000
                                  700,000

       U n i ts , M il li o n s
                                  600,000
                                  500,000
                                  400,000
                                  300,000
                                  200,000
                                  100,000
                                      -
                                            00

                                                 01

                                                      02

                                                           03

                                                                04

                                                                      05

                                                                            06

                                                                                 07

                                                                                        08

                                                                                              09

                                                                                                   10

                                                                                                        11

                                                                                                              12
                                        CY

                                             CY

                                                  CY

                                                       CY

                                                              CY

                                                                   CY

                                                                         CY

                                                                               CY

                                                                                      CY

                                                                                           CY

                                                                                                 CY

                                                                                                      CY

                                                                                                           CY
                                                                                     Source: Seagate Market Research




 Rapid overall HDD unit growth will continue into the
 foreseeable future
 More than 1.5X increase in units shipped in 2012
 compared to 2007


                                                           Andrei Khurshudov, 2007
Strong Link Between Information Growth and
             Storage Produced
• Internet
• Blogs
• Movies
• TV
• Music
• Maps
• Databases
• Archives
                                                   New Storage
                      New Data
• Business
• Legal
• Science
• Diaries
• Art
• Gaming
• Literature
• Noise
• Etc.

               Balance is required!
               Data storage technology underpins information growth
                              Andrei Khurshudov, 2007
Estimated Total PB’s Capacity Shipped
                                                  T otal PB's shippe d Proje ction                       y = 7872.3e 0.3679x
                                                                                                            R 2 = 0.9883
                        500,000

                        450,000

                        400,000
                                                                       Exponential growth
                        350,000
     Total PB 's ship




                        300,000

                        250,000

                        200,000

                        150,000

                        100,000

                         50,000

                            -
                                  2002   2003   2004    2005    2006    2007     2008     2009    2010      2011      2012

                                                                        Ye ar
                                                                                        Source: Seagate Market Research



  Information growth trend is indeed exponential!
  Overall information growth will scale with the HDD capacity growth
     It is estimated that over 90% of all new information produced in the world is being stored on
     magnetic media, most of it on hard disk drives (Google)
  Shipped capacity doubles every 30 months
  Over 1M PB of storage will be produced between 2008-12
                                                       Andrei Khurshudov, 2007
Long-term Storage Growth Projection
                                              Long-term storage growth projection
                                                                                                      Alotabyte?
                                                                                                               !!!
                       1,000,000,000,000
                        100,000,000,000
  Total PB PB shippe
     Total Shipped


                         10,000,000,000
                                                                                                              Yottabyte
                          1,000,000,000
                            100,000,000
                             10,000,000
                                                                                                              Zettabyte
                              1,000,000
                                100,000
                                 10,000
                                                                                                              Exabyte
                                  1,000
                                    100
                                     10
                                                                                                              Petabyte
                                      1
                                      2000   2005   2010       2015     2020     2025   2030   2035    2040    2045      2050

                                                                                 Year                         Andrei Khurshudov




     Exponential growth in storage capacity will
        enable the information avalanche!
                                                           Andrei Khurshudov, 2007
Definitions of reliability
Reliability is the probability of performing required functions for a
specified time under the stated operational conditions
For HDD:
   Required functions include storing and accessing data at the specified high
   data rate and with specified power consumption, acoustic noise, start-up
   time, etc.
   Specified time is the service life, which is typically 3 to 10 years.
   Stated operational conditions are those specified by the HDD
   specification (temperature, humidity, shock, vibration, etc.)
Weibull reliability model :
  Describes the “weakest link” in a product
  Treats system as a series of components each having finite
  reliability:
               R1        R2                                       Rn
                                    HDD Reliability

                                                                 Etc.
                                                          Code
                        Motor            PCBA
              HDI

                      HDD fails if any one component fails!
                                                                 R = R1*R2*R3*…Rn
                                Andrei Khurshudov, 2007
HDD Reliability Trends
                                       Manufacturer’s HDD MTBF Specifications

                                              From: Ed Grochowski, IBM
                                             From: Ed Grochowski, IBM


                                                                                      • MTBF indicates,
                                                                                      on average, how
                                                                                      many hours a
                                                                                      product is expected
                                                                                      to operate before
                                                                                      failures.
                                                                                      • MTBF = Total
      The Ultimate Battle
                                                                                      Product
 Reliability vs. Storage Density                                                      Operational Time /
                                                                                      Number of Failures
      Reliability vs. Cost
                                   Current typical MTBF numbers (by product class):
  Reliability vs. Performance
                                         Server: 1,400,000 hours
Reliability vs. Development Time         Desktop: 700,000 hours
                                         Mobile: 400,000 hours
  Reliability vs. Environment
               …


                          Reliability keeps increasing with time in spite of design
                            complexities and more stringent qualification test
                                               requirements
                                                    Andrei Khurshudov, 2007
HDD Reliability Hierarchy
      Involvement                                         Dealing with…

Customer perception of reliability                             Limited statistics


                                                     Closing gap between expected
Reliability in User Environment
                                                          reliability and reality

                                                          The last line of defense.
  Manufacturing for Reliability
                                                      Balancing quality against cost

                                                       Advanced test techniques and
 Product reliability qualification
                                                          failure modes analysis


                                                        Engineering and Technology
      Design for Reliability
                                                                 Principles


  Reliability Physics & Theory                          Fundamental laws of nature



                     HDD reliability is built upon Tribology !
                                     Andrei Khurshudov, 2007
A Perspective on HDD Reliability
                  Cumulative Failure / Repair / Return rates (after 3-4 years)

                                    Laptop com puter
Refrigerator: side-by-side, w icem
                             ith  aker and dispenser
                                        Rider m er
                                                ow
                                  Desktop com   puter
                                                                                                                                         When Compared to
                                         Law tractor
                                             n
                      Washing machine (front-loading)
                                                                                                                                         many other products,
                                 Self-propelled m er
                                                  ow
                            Vacuum cleaner (canister)
                                                                                                                                         HDD reliability looks
                        Washing m achine (top-loading)
                                           Dishw asher
                                                                                                                                         very high
                                            Gas range
    Refrigerator: top- and bottom-freezer, w icem
                                            /     aker

                                                                                                                                         Average 3-4 year
                                 W oven (electric)
                                   all
                                  Push m er (gas)
                                          ow

                                                                                                                                         cumulative repair
                     Microwave oven (over-the-range)
                                       Cooktop (gas)
                                         Clothes dryer
                                                                                                                                         rate for CE products
                              Average for CE products
                             Vacuum cleaner (upright)
                                                                                                                                         is 15%
                                    Cam corder (digital)
    Refrigerator: top- and bottom-freezer, no icemaker
                                                                                                                                         HDD is a component,
                                     Cooktop (electric)
                                      Range (electric)
                                                                                                                                         not a product
                                        Digital cam era
                         TV: 30- to 36-inch direct view
                         TV: 25- to 27-inch direct view
                                         Proton rocket
                                                  HDD
                                  M edical Pacem akers
                                  Sony PS3 (w H
                                              ith DD)

                                                                                                                                     %
                                                           0   5     10     15     20     25    30     35     40    45     50


   Source: Consumer Reports National Research Center, 2006 Product Reliability Survey; http://en.wikipedia.org/wiki/Proton_rocket;
   www.seagate.com; http://www.medscape.com/viewarticle/536755

                                                                              Andrei Khurshudov, 2007
The Actual Cost of Unreliability
If the company experiences a major loss of data then
   60% of companies that lose their data will shut down within 6 months of the
   disaster (source: Bostoncomputing.net))
                     Bostoncomputing.net
   72% of businesses that suffer major data loss disappear within 24 months
   (Source: Realty Times)

   93% of companies that lost their data center for 10 days or more due to a
   disaster filed for bankruptcy within one year of the disaster (source:
   Bostoncomputing.net)
   Bostoncomputing.net)

Recreating data from scratch is estimated to cost between $2000
and $8000 per MB (Source: Realty Times)
Of those companies participating in the 2001 Cost of Downtime
Survey (Source: 2001 Cost of Downtime Survey Results):
   8% said it would cost their companies more than $1 million per hour
   18% said each hour would cost between $251K and $1 million
   28% said each hour would cost between $51K and $250K
   46% said each hour of downtime would cost their companies up to $50k



                                Andrei Khurshudov, 2007
Aggravating Aspects of Data Loss
40% of Small and Medium Sized Businesses do not back up their data (Source: Realty
Times)

40 - 50% of all backups are not fully recoverable (Source: Realty Times)
34% of companies fail to test their tape backups, and
    of those that do, 77% have found tape back-up failures (source: Bostoncomputing.net))
                                                                         Bostoncomputing.net

quot;More than 109,000 TBs of unique enterprise PC data are not being regularly
backed up“ (IDC)

A national Harris Interactive survey reveals (Source: Realty Times):
         Only 25% of users frequently back up digital files, even when 85 percent of
         computer users say they are very concerned about losing important digital data
         37% of the survey's respondents admitted to backing up their files less than once
         per month
         9% admitted they have never backed up their files
         More than 22% said backing up information is on their to-do list, but they
         seldom do it




                                     Andrei Khurshudov, 2007
What do drives fail for?
                                    Generic HDD failure mode pareto
                                          Write abort
                   High-fly write

                                                                                    • Up to 40%
                                                                  NTF
                                                                 CND
         Scratch                                                                    • System-dependent



         TA




Head degradation
                                                                                    • Up to 30%
                                                                                    • System-dependent,
        Grown defect
                                                                                    personnel-dependent,
                        Motor
                                                                                    procedure-dependent,
                                                            Mishandling
                                                             Handling      damage
                                    PCB
                                                                                    etc.




                           Observation:
                                    Tribology is responsible for many failure modes !

                                                        Andrei Khurshudov, 2007
Tribology inside HDD
                       Connectors




                                                    FDB Motor



                                                 Head-Disk Interface

                                                   Ramp (friction and wear)
Pivot Bearing
                                                        Screws
                                               (wear and torque retention)




There are multiple ways in which tribology impacts HDD reliability
                     Andrei Khurshudov, 2007
The Role of Tribology in HDD Reliability

It is estimated that 15% to 35% of all HDD failures are
linked to Tribology (25% on average)
  Improving tribological robustness enhances overall disk drive
  reliability

Major known failure modes related to tribological issues:
  Scratch (on both head and media; with or w/out particles)
  Thermal erasure (disk) and head degradation
  New defects
  Weak write / read
  Crash
  Failure of some other moving parts
  Etc.

                       Andrei Khurshudov, 2007
Future Improvement Opportunities

HDD reliability:
  Number of drives that will not fail between 2008 and 2012 per
  every 0.1% AFR improvement: ~ 3,000,000
  Amount of stored information that will not be lost/impacted
  between 2008 and 2012 per every 0.1% AFR improvement:
  ~ 1,000,000 TB (or 1 EB)
Tribology:
  Number of drives that will not fail between 2008 and 2012 due to
  Tribological problems per every 0.1% AFR improvement: ~
  750,000
  Amount of stored information that will not be lost/impacted
  between 2008 and 2012 due to Tribological problems per every
  0.1% AFR improvement: ~ 250,000 TB = 250 PB

                         Andrei Khurshudov, 2007
Is this worth the effort?
Petabytes in use:
  The “American Memory” project is one of the largest digitized archives of U.S.
  history, with more than 7.5 million digital records from 100 collections of
  manuscripts, books, maps, films, sound recordings and photographs. The total
  size of the project is 0.008 Petabytes [Wired]
  As of November 2006, eBay had 2 Petabytes of data [Wikipedia]
                                                             [Wikipedia
  Jefferson National Accelerator Facility has a 2 Petabyte storage farm used to
  collect data from experiments on the particle accelerator [Wikipedia]
                                                                    [Wikipedia
  RapidShare in 2007 had 3.5 Petabytes of hard-disk storage [Wikipedia] [Wikipedia
  The San Diego Supercomputer Center (SDSC) in the USA has a 1-Petabyte hard
  disk store and a 6-Petabyte robotic tape store [Wikipedia]
                                                  [Wikipedia
  Microsoft stores on 900 servers a total of about 14 Petabytes. These are mostly
  imagery for Microsoft's digital model planet, Virtual Earth [Wikipedia]
  15 Petabytes of data will be generated each year in particle physics experiments
  using CERN’s Large Hadron Collider, due to be launched in May 2008 [Wikipedia]   [Wikipedia
  The total storage capacity needed for the above data is ~ 44 PB
  A failure rate reduction of 0.005% over the next 5 years is required to
  cover the above storage capacity needs

                                   Andrei Khurshudov, 2007
Future Scenario
Exponential growth of data over time
(information avalanche)
Lower cost of data storage per GB
Many more disk drives required to
accommodate all of the new data and backup
Continually increasing reliability of disk drives
Nevertheless, more total failures (in absolute
terms) unless HDD reliability increases on a
faster rate than the drive unit growth


                   Andrei Khurshudov, 2007
Conclusions

Data storage capacity growth enables overall
information growth
Reliability of data storage devices is a key element in this
growth
   Unreliability is extremely costly
Even small improvements in reliability will have huge
impact on the amount of information preserved in the
future
Tribology is, and will remain, a major enabler of the
future information growth
   Relative contribution of Tribology to HDD unreliability is on
   the order of 25%

                          Andrei Khurshudov, 2007
References
“The Innovator’s Dilemma” by Clayton M. Christensen
Google: Failure Trends in a Large Disk Drive Population, E. Pinheiro, W.-D. Weber and
L. Andr´e Barroso, FAST 2007
Wired: http://www.wired.com/science/discoveries/news/2002/10/55509
Wikipedia on Petabytes: http://en.wikipedia.org/wiki/Petabyte
Consumer Reports National Research Center, 2006 Product Reliability Survey:
http://www.squaretrade.com/htm/pop/lm_failureRates.html
Proton rocket launcher: http://en.wikipedia.org/wiki/Proton_rocket
HDD specifications: www.seagate.com
Medical pacemaker’s reliability: http://www.medscape.com/viewarticle/536755
2001 Cost of Downtime Survey Results: http://www.datadepositbox.com/media/data-
loss-statistics.asp
BostonComputing.net:
http://www.bostoncomputing.net/consultation/databackup/statistics
IDC: IDC analyst Fred Broussard, PC Backup and Higher Prioritization for the
Enterprise and Consumer, July 2002




                                Andrei Khurshudov, 2007
Thank you!




  Andrei Khurshudov, 2007

Weitere ähnliche Inhalte

Andere mochten auch

Prof William Kosar: Letters of Credit as a Payment Method
Prof William Kosar: Letters of Credit as a Payment MethodProf William Kosar: Letters of Credit as a Payment Method
Prof William Kosar: Letters of Credit as a Payment MethodWilliam Kosar
 
TTN Sports Pingpong Training 1
TTN Sports Pingpong Training 1TTN Sports Pingpong Training 1
TTN Sports Pingpong Training 1Duong Thinh
 
Bob Dylan Y Serrat En Hebreo
Bob Dylan Y Serrat En HebreoBob Dylan Y Serrat En Hebreo
Bob Dylan Y Serrat En Hebreosilvia shapiro
 
Rims Metal and Mining Session talk by F+C Oboni, Riskope
Rims Metal and Mining Session talk by F+C Oboni, RiskopeRims Metal and Mining Session talk by F+C Oboni, Riskope
Rims Metal and Mining Session talk by F+C Oboni, RiskopeOboni Riskope Associates Inc.
 
Uye Cekim Degerlendirmesi
Uye Cekim DegerlendirmesiUye Cekim Degerlendirmesi
Uye Cekim DegerlendirmesiSamet Tuna
 
Web 2.0, Nederlandse Spoorwegen En Proefstation Leiden
Web 2.0, Nederlandse Spoorwegen En Proefstation LeidenWeb 2.0, Nederlandse Spoorwegen En Proefstation Leiden
Web 2.0, Nederlandse Spoorwegen En Proefstation LeidenCoenDirkx
 
Responders: Respond-Report-Review
Responders: Respond-Report-ReviewResponders: Respond-Report-Review
Responders: Respond-Report-Reviewfrewsmhuffman
 
Social Media and Fundraisng - are you prepared?
Social Media and Fundraisng - are you prepared? Social Media and Fundraisng - are you prepared?
Social Media and Fundraisng - are you prepared? Noesium Consulting
 
Accurate biochemical knowledge starting with precise structure-based criteria...
Accurate biochemical knowledge starting with precise structure-based criteria...Accurate biochemical knowledge starting with precise structure-based criteria...
Accurate biochemical knowledge starting with precise structure-based criteria...Michel Dumontier
 
Combining Quantitative & Qualitative Data in a Single Large scale User Resear...
Combining Quantitative & Qualitative Data in a Single Large scale User Resear...Combining Quantitative & Qualitative Data in a Single Large scale User Resear...
Combining Quantitative & Qualitative Data in a Single Large scale User Resear...UserZoom
 
Intro to Twitter for Companies
Intro to Twitter for CompaniesIntro to Twitter for Companies
Intro to Twitter for CompaniesNoesium Consulting
 
BigScrum - Scaling Teams to Programs
BigScrum - Scaling Teams to ProgramsBigScrum - Scaling Teams to Programs
BigScrum - Scaling Teams to ProgramsThinkLouder
 

Andere mochten auch (19)

Prof William Kosar: Letters of Credit as a Payment Method
Prof William Kosar: Letters of Credit as a Payment MethodProf William Kosar: Letters of Credit as a Payment Method
Prof William Kosar: Letters of Credit as a Payment Method
 
Report z Vyškov
Report z VyškovReport z Vyškov
Report z Vyškov
 
Social Media Summit
Social Media SummitSocial Media Summit
Social Media Summit
 
D4 I Intro Web
D4 I Intro WebD4 I Intro Web
D4 I Intro Web
 
TTN Sports Pingpong Training 1
TTN Sports Pingpong Training 1TTN Sports Pingpong Training 1
TTN Sports Pingpong Training 1
 
Bob Dylan Y Serrat En Hebreo
Bob Dylan Y Serrat En HebreoBob Dylan Y Serrat En Hebreo
Bob Dylan Y Serrat En Hebreo
 
Rims Metal and Mining Session talk by F+C Oboni, Riskope
Rims Metal and Mining Session talk by F+C Oboni, RiskopeRims Metal and Mining Session talk by F+C Oboni, Riskope
Rims Metal and Mining Session talk by F+C Oboni, Riskope
 
Uye Cekim Degerlendirmesi
Uye Cekim DegerlendirmesiUye Cekim Degerlendirmesi
Uye Cekim Degerlendirmesi
 
Web 2.0, Nederlandse Spoorwegen En Proefstation Leiden
Web 2.0, Nederlandse Spoorwegen En Proefstation LeidenWeb 2.0, Nederlandse Spoorwegen En Proefstation Leiden
Web 2.0, Nederlandse Spoorwegen En Proefstation Leiden
 
Responders: Respond-Report-Review
Responders: Respond-Report-ReviewResponders: Respond-Report-Review
Responders: Respond-Report-Review
 
Social Media and Fundraisng - are you prepared?
Social Media and Fundraisng - are you prepared? Social Media and Fundraisng - are you prepared?
Social Media and Fundraisng - are you prepared?
 
Earth Day
Earth DayEarth Day
Earth Day
 
Accurate biochemical knowledge starting with precise structure-based criteria...
Accurate biochemical knowledge starting with precise structure-based criteria...Accurate biochemical knowledge starting with precise structure-based criteria...
Accurate biochemical knowledge starting with precise structure-based criteria...
 
Combining Quantitative & Qualitative Data in a Single Large scale User Resear...
Combining Quantitative & Qualitative Data in a Single Large scale User Resear...Combining Quantitative & Qualitative Data in a Single Large scale User Resear...
Combining Quantitative & Qualitative Data in a Single Large scale User Resear...
 
G4
G4G4
G4
 
Picnik
PicnikPicnik
Picnik
 
Free Software
Free SoftwareFree Software
Free Software
 
Intro to Twitter for Companies
Intro to Twitter for CompaniesIntro to Twitter for Companies
Intro to Twitter for Companies
 
BigScrum - Scaling Teams to Programs
BigScrum - Scaling Teams to ProgramsBigScrum - Scaling Teams to Programs
BigScrum - Scaling Teams to Programs
 

Ähnlich wie Future Information Growth And Storage Device Reliability 2007

sample ppt- Subject Introduction (1).pptx
sample ppt- Subject Introduction (1).pptxsample ppt- Subject Introduction (1).pptx
sample ppt- Subject Introduction (1).pptxShubhamRai938546
 
Hard disk & Optical disk (college group project)
Hard disk & Optical disk (college group project)Hard disk & Optical disk (college group project)
Hard disk & Optical disk (college group project)Vshal_Rai
 
Cd &-dvd-by-aaron-rinaca-mike-ferris-mike-burker-steve-mathieu-2001-spr
Cd &-dvd-by-aaron-rinaca-mike-ferris-mike-burker-steve-mathieu-2001-sprCd &-dvd-by-aaron-rinaca-mike-ferris-mike-burker-steve-mathieu-2001-spr
Cd &-dvd-by-aaron-rinaca-mike-ferris-mike-burker-steve-mathieu-2001-sprVarun Kumar
 
Blu Ray Disc Seminar
Blu Ray Disc SeminarBlu Ray Disc Seminar
Blu Ray Disc SeminarRajesh Kumar
 
Compare CD vs DVD
Compare CD vs DVDCompare CD vs DVD
Compare CD vs DVDPratik Vyas
 
Blu ray disk seminar
Blu ray disk seminarBlu ray disk seminar
Blu ray disk seminarPranay Raj
 
7 cs102 dvd report (2)
7 cs102 dvd report (2)7 cs102 dvd report (2)
7 cs102 dvd report (2)Tweetie Sabado
 
Removeable Storage
Removeable StorageRemoveable Storage
Removeable StorageJoel May
 
Erika Bonati Sections 1.3 Project 3
Erika Bonati Sections 1.3 Project 3Erika Bonati Sections 1.3 Project 3
Erika Bonati Sections 1.3 Project 3erikabonati
 
Disco m disc
Disco m discDisco m disc
Disco m discjasl92
 
External memory
External memoryExternal memory
External memoryriddhishg
 
Disco m disc
Disco m discDisco m disc
Disco m discjasl92
 

Ähnlich wie Future Information Growth And Storage Device Reliability 2007 (20)

sample ppt- Subject Introduction (1).pptx
sample ppt- Subject Introduction (1).pptxsample ppt- Subject Introduction (1).pptx
sample ppt- Subject Introduction (1).pptx
 
shubham rai chs.pptx
shubham rai chs.pptxshubham rai chs.pptx
shubham rai chs.pptx
 
Hard disk & Optical disk (college group project)
Hard disk & Optical disk (college group project)Hard disk & Optical disk (college group project)
Hard disk & Optical disk (college group project)
 
Other data storage devices v3
Other data storage devices v3Other data storage devices v3
Other data storage devices v3
 
shubham cs.pptx
shubham cs.pptxshubham cs.pptx
shubham cs.pptx
 
Cd &-dvd-by-aaron-rinaca-mike-ferris-mike-burker-steve-mathieu-2001-spr
Cd &-dvd-by-aaron-rinaca-mike-ferris-mike-burker-steve-mathieu-2001-sprCd &-dvd-by-aaron-rinaca-mike-ferris-mike-burker-steve-mathieu-2001-spr
Cd &-dvd-by-aaron-rinaca-mike-ferris-mike-burker-steve-mathieu-2001-spr
 
Blu Ray Disc Seminar
Blu Ray Disc SeminarBlu Ray Disc Seminar
Blu Ray Disc Seminar
 
Cd rom
Cd romCd rom
Cd rom
 
Compare CD vs DVD
Compare CD vs DVDCompare CD vs DVD
Compare CD vs DVD
 
HVD
HVDHVD
HVD
 
Presented by
Presented byPresented by
Presented by
 
Blu ray disk seminar
Blu ray disk seminarBlu ray disk seminar
Blu ray disk seminar
 
Secondary Storage
Secondary StorageSecondary Storage
Secondary Storage
 
storage device
storage device storage device
storage device
 
7 cs102 dvd report (2)
7 cs102 dvd report (2)7 cs102 dvd report (2)
7 cs102 dvd report (2)
 
Removeable Storage
Removeable StorageRemoveable Storage
Removeable Storage
 
Erika Bonati Sections 1.3 Project 3
Erika Bonati Sections 1.3 Project 3Erika Bonati Sections 1.3 Project 3
Erika Bonati Sections 1.3 Project 3
 
Disco m disc
Disco m discDisco m disc
Disco m disc
 
External memory
External memoryExternal memory
External memory
 
Disco m disc
Disco m discDisco m disc
Disco m disc
 

Mehr von Andrei Khurshudov

Hyper-Converged Infrastructure: Big Data and IoT opportunities and challenges...
Hyper-Converged Infrastructure: Big Data and IoT opportunities and challenges...Hyper-Converged Infrastructure: Big Data and IoT opportunities and challenges...
Hyper-Converged Infrastructure: Big Data and IoT opportunities and challenges...Andrei Khurshudov
 
Short introduction to Big Data Analytics, the Internet of Things, and their s...
Short introduction to Big Data Analytics, the Internet of Things, and their s...Short introduction to Big Data Analytics, the Internet of Things, and their s...
Short introduction to Big Data Analytics, the Internet of Things, and their s...Andrei Khurshudov
 
Health monitoring & predictive analytics to lower the TCO in a datacenter
Health monitoring & predictive analytics to lower the TCO in a datacenterHealth monitoring & predictive analytics to lower the TCO in a datacenter
Health monitoring & predictive analytics to lower the TCO in a datacenterAndrei Khurshudov
 
clusterstor-hadoop-data-sheet
clusterstor-hadoop-data-sheetclusterstor-hadoop-data-sheet
clusterstor-hadoop-data-sheetAndrei Khurshudov
 
Reliability Of Solid State Drives 2008
Reliability Of Solid State Drives 2008Reliability Of Solid State Drives 2008
Reliability Of Solid State Drives 2008Andrei Khurshudov
 

Mehr von Andrei Khurshudov (9)

Hyper-Converged Infrastructure: Big Data and IoT opportunities and challenges...
Hyper-Converged Infrastructure: Big Data and IoT opportunities and challenges...Hyper-Converged Infrastructure: Big Data and IoT opportunities and challenges...
Hyper-Converged Infrastructure: Big Data and IoT opportunities and challenges...
 
Short introduction to Big Data Analytics, the Internet of Things, and their s...
Short introduction to Big Data Analytics, the Internet of Things, and their s...Short introduction to Big Data Analytics, the Internet of Things, and their s...
Short introduction to Big Data Analytics, the Internet of Things, and their s...
 
Seagate_1
Seagate_1Seagate_1
Seagate_1
 
Health monitoring & predictive analytics to lower the TCO in a datacenter
Health monitoring & predictive analytics to lower the TCO in a datacenterHealth monitoring & predictive analytics to lower the TCO in a datacenter
Health monitoring & predictive analytics to lower the TCO in a datacenter
 
Using Big Data Analytics
Using Big Data AnalyticsUsing Big Data Analytics
Using Big Data Analytics
 
Presentation_Final
Presentation_FinalPresentation_Final
Presentation_Final
 
clusterstor-hadoop-data-sheet
clusterstor-hadoop-data-sheetclusterstor-hadoop-data-sheet
clusterstor-hadoop-data-sheet
 
Long Term Data Storage 2007
Long Term Data Storage 2007Long Term Data Storage 2007
Long Term Data Storage 2007
 
Reliability Of Solid State Drives 2008
Reliability Of Solid State Drives 2008Reliability Of Solid State Drives 2008
Reliability Of Solid State Drives 2008
 

Kürzlich hochgeladen

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
 
A Framework for Development in the AI Age
A Framework for Development in the AI AgeA Framework for Development in the AI Age
A Framework for Development in the AI AgeCprime
 
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...Scott Andery
 
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
 
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
 
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
 
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...Wes McKinney
 
Potential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and InsightsPotential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and InsightsRavi Sanghani
 
Manual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance AuditManual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance AuditSkynet Technologies
 
So einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdfSo einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdfpanagenda
 
DevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platformsDevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platformsSergiu Bodiu
 
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
 
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
 
Moving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfMoving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfLoriGlavin3
 
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
 
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
 
Testing tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examplesTesting tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examplesKari Kakkonen
 
Data governance with Unity Catalog Presentation
Data governance with Unity Catalog PresentationData governance with Unity Catalog Presentation
Data governance with Unity Catalog PresentationKnoldus Inc.
 
What is DBT - The Ultimate Data Build Tool.pdf
What is DBT - The Ultimate Data Build Tool.pdfWhat is DBT - The Ultimate Data Build Tool.pdf
What is DBT - The Ultimate Data Build Tool.pdfMounikaPolabathina
 
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
 

Kürzlich hochgeladen (20)

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
 
A Framework for Development in the AI Age
A Framework for Development in the AI AgeA Framework for Development in the AI Age
A Framework for Development in the AI Age
 
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...
Enhancing User Experience - Exploring the Latest Features of Tallyman Axis Lo...
 
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
 
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
 
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
 
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
 
Potential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and InsightsPotential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and Insights
 
Manual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance AuditManual 508 Accessibility Compliance Audit
Manual 508 Accessibility Compliance Audit
 
So einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdfSo einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdf
 
DevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platformsDevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platforms
 
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
 
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
 
Moving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfMoving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.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
 
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
 
Testing tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examplesTesting tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examples
 
Data governance with Unity Catalog Presentation
Data governance with Unity Catalog PresentationData governance with Unity Catalog Presentation
Data governance with Unity Catalog Presentation
 
What is DBT - The Ultimate Data Build Tool.pdf
What is DBT - The Ultimate Data Build Tool.pdfWhat is DBT - The Ultimate Data Build Tool.pdf
What is DBT - The Ultimate Data Build Tool.pdf
 
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
 

Future Information Growth And Storage Device Reliability 2007

  • 1. Future Information Growth AND Storage Device Reliability Andrei Khurshudov Seagate Technology 2007
  • 2. The History of Data Storage • Storage media: charcoal and dirt on stone • Data type: analog (image) • Storage life: >17,000 years (in a sealed dry ‘Diamond Sutra’ (the world’s earliest complete survival of a dated printed book), cave) AD 868 Storage media: ink on paper Data type: analog (images, characters) Storage life: >1,100 years (sealed in a cave) Andrei Khurshudov, 2007
  • 3. The History of Computer Data Storage 1.8” Perpendicular 2005 5.25” drive 2.5” drive 1.8” drive RAMAC Hard disk drive 3340 Winchester, 1980 1991 1988 1956 1962 Hybrid Jazz Zip The floppy 3.5” drive SSD Magnetic drum 1983 Blue-ray/HD DVD Don’t know how to sell more storage… DVD CD ROM Direct access to data Magnetic Tape Holographic Disk disk 1980 CD/DVD Holographic Sequential access to data Need more storage! Punch cards Compact Cassette Magnetic tape Punched tape 1940 1950 1960 1970 1980 1990 2000 2010 2020 quot;Do not fold, spindle or mutilate” Andrei Khurshudov, 2007
  • 4. The First HDD is Born • Stands for quot;Random Access Method of Accounting and Controlquot; • Born: 1956 • Capacity: 5 MB • Disk diameter: 24” • Recording surfaces: 100 • Tracks/surface: 100 • RPM: 1200 • Weight: >1 ton • Cost: leased for $3,200 per month “While the storage capacity of the drive could have been increased above five megabytes, the marketing department at IBM was against a larger capacity drive because they didn't know how to sell a product with more storage (source: Currie Munce, VP, IBM Research) Andrei Khurshudov, 2007
  • 5. Modern Disk Drive About 50 years old Runs faster with every year… Mass-produced electro-mechanical device 2006 total industry output >400M drives Utilizes principles of magnetic recording Most recent products utilize PMR Relies on a flying magnetic element Typical mechanical separation ~5-10 nm Available in several standard form factors 1”, 1.8”, 2.5”, 3.5” Designed for several distinct markets Desktop, Enterprise, Mobile, HH, CE Uses various computer interfaces PATA, SATA, SAS, SCSI, FCAL Historically high data density growth rate CAGR of 30% to 50% over the last decades Experiences constant cost pressure Cost of GB is under $0.5 and falling Always under attack from disruptive Destroys or assimilates competition for 50 technologies years Continually expands into new markets Most recent: CE, automotive, archival Highly competitive industry Darwinian principles in accelerated action* Industry share leader: Seagate ~40% of the total market share * “The Innovator’s Dilemma” by Clayton M. Christensen Innovator’ Dilemma” Andrei Khurshudov, 2007
  • 6. Disk Drive Industry Trends 0.85” drive Source: PC World, The Hard Drive Turns 50 Source: Coughlin Associates Bear Stearns Technology Conference, 2006 Bear Stearns Technology Conference, 2006 Ed Grochowski, IBM Ed Grochowski, IBM Drives get denser, smaller, faster, and cheaper Reliability becomes increasingly difficult Andrei Khurshudov, 2007
  • 7. Yesterday, Today, and Tomorrow Tomorrow Yesterday Today There’s plenty of room at the bottom! Andrei Khurshudov, 2007
  • 8. Estimated Number of Units Shipped 900,000 800,000 700,000 U n i ts , M il li o n s 600,000 500,000 400,000 300,000 200,000 100,000 - 00 01 02 03 04 05 06 07 08 09 10 11 12 CY CY CY CY CY CY CY CY CY CY CY CY CY Source: Seagate Market Research Rapid overall HDD unit growth will continue into the foreseeable future More than 1.5X increase in units shipped in 2012 compared to 2007 Andrei Khurshudov, 2007
  • 9. Strong Link Between Information Growth and Storage Produced • Internet • Blogs • Movies • TV • Music • Maps • Databases • Archives New Storage New Data • Business • Legal • Science • Diaries • Art • Gaming • Literature • Noise • Etc. Balance is required! Data storage technology underpins information growth Andrei Khurshudov, 2007
  • 10. Estimated Total PB’s Capacity Shipped T otal PB's shippe d Proje ction y = 7872.3e 0.3679x R 2 = 0.9883 500,000 450,000 400,000 Exponential growth 350,000 Total PB 's ship 300,000 250,000 200,000 150,000 100,000 50,000 - 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Ye ar Source: Seagate Market Research Information growth trend is indeed exponential! Overall information growth will scale with the HDD capacity growth It is estimated that over 90% of all new information produced in the world is being stored on magnetic media, most of it on hard disk drives (Google) Shipped capacity doubles every 30 months Over 1M PB of storage will be produced between 2008-12 Andrei Khurshudov, 2007
  • 11. Long-term Storage Growth Projection Long-term storage growth projection Alotabyte? !!! 1,000,000,000,000 100,000,000,000 Total PB PB shippe Total Shipped 10,000,000,000 Yottabyte 1,000,000,000 100,000,000 10,000,000 Zettabyte 1,000,000 100,000 10,000 Exabyte 1,000 100 10 Petabyte 1 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Year Andrei Khurshudov Exponential growth in storage capacity will enable the information avalanche! Andrei Khurshudov, 2007
  • 12. Definitions of reliability Reliability is the probability of performing required functions for a specified time under the stated operational conditions For HDD: Required functions include storing and accessing data at the specified high data rate and with specified power consumption, acoustic noise, start-up time, etc. Specified time is the service life, which is typically 3 to 10 years. Stated operational conditions are those specified by the HDD specification (temperature, humidity, shock, vibration, etc.) Weibull reliability model : Describes the “weakest link” in a product Treats system as a series of components each having finite reliability: R1 R2 Rn HDD Reliability Etc. Code Motor PCBA HDI HDD fails if any one component fails! R = R1*R2*R3*…Rn Andrei Khurshudov, 2007
  • 13. HDD Reliability Trends Manufacturer’s HDD MTBF Specifications From: Ed Grochowski, IBM From: Ed Grochowski, IBM • MTBF indicates, on average, how many hours a product is expected to operate before failures. • MTBF = Total The Ultimate Battle Product Reliability vs. Storage Density Operational Time / Number of Failures Reliability vs. Cost Current typical MTBF numbers (by product class): Reliability vs. Performance Server: 1,400,000 hours Reliability vs. Development Time Desktop: 700,000 hours Mobile: 400,000 hours Reliability vs. Environment … Reliability keeps increasing with time in spite of design complexities and more stringent qualification test requirements Andrei Khurshudov, 2007
  • 14. HDD Reliability Hierarchy Involvement Dealing with… Customer perception of reliability Limited statistics Closing gap between expected Reliability in User Environment reliability and reality The last line of defense. Manufacturing for Reliability Balancing quality against cost Advanced test techniques and Product reliability qualification failure modes analysis Engineering and Technology Design for Reliability Principles Reliability Physics & Theory Fundamental laws of nature HDD reliability is built upon Tribology ! Andrei Khurshudov, 2007
  • 15. A Perspective on HDD Reliability Cumulative Failure / Repair / Return rates (after 3-4 years) Laptop com puter Refrigerator: side-by-side, w icem ith aker and dispenser Rider m er ow Desktop com puter When Compared to Law tractor n Washing machine (front-loading) many other products, Self-propelled m er ow Vacuum cleaner (canister) HDD reliability looks Washing m achine (top-loading) Dishw asher very high Gas range Refrigerator: top- and bottom-freezer, w icem / aker Average 3-4 year W oven (electric) all Push m er (gas) ow cumulative repair Microwave oven (over-the-range) Cooktop (gas) Clothes dryer rate for CE products Average for CE products Vacuum cleaner (upright) is 15% Cam corder (digital) Refrigerator: top- and bottom-freezer, no icemaker HDD is a component, Cooktop (electric) Range (electric) not a product Digital cam era TV: 30- to 36-inch direct view TV: 25- to 27-inch direct view Proton rocket HDD M edical Pacem akers Sony PS3 (w H ith DD) % 0 5 10 15 20 25 30 35 40 45 50 Source: Consumer Reports National Research Center, 2006 Product Reliability Survey; http://en.wikipedia.org/wiki/Proton_rocket; www.seagate.com; http://www.medscape.com/viewarticle/536755 Andrei Khurshudov, 2007
  • 16. The Actual Cost of Unreliability If the company experiences a major loss of data then 60% of companies that lose their data will shut down within 6 months of the disaster (source: Bostoncomputing.net)) Bostoncomputing.net 72% of businesses that suffer major data loss disappear within 24 months (Source: Realty Times) 93% of companies that lost their data center for 10 days or more due to a disaster filed for bankruptcy within one year of the disaster (source: Bostoncomputing.net) Bostoncomputing.net) Recreating data from scratch is estimated to cost between $2000 and $8000 per MB (Source: Realty Times) Of those companies participating in the 2001 Cost of Downtime Survey (Source: 2001 Cost of Downtime Survey Results): 8% said it would cost their companies more than $1 million per hour 18% said each hour would cost between $251K and $1 million 28% said each hour would cost between $51K and $250K 46% said each hour of downtime would cost their companies up to $50k Andrei Khurshudov, 2007
  • 17. Aggravating Aspects of Data Loss 40% of Small and Medium Sized Businesses do not back up their data (Source: Realty Times) 40 - 50% of all backups are not fully recoverable (Source: Realty Times) 34% of companies fail to test their tape backups, and of those that do, 77% have found tape back-up failures (source: Bostoncomputing.net)) Bostoncomputing.net quot;More than 109,000 TBs of unique enterprise PC data are not being regularly backed up“ (IDC) A national Harris Interactive survey reveals (Source: Realty Times): Only 25% of users frequently back up digital files, even when 85 percent of computer users say they are very concerned about losing important digital data 37% of the survey's respondents admitted to backing up their files less than once per month 9% admitted they have never backed up their files More than 22% said backing up information is on their to-do list, but they seldom do it Andrei Khurshudov, 2007
  • 18. What do drives fail for? Generic HDD failure mode pareto Write abort High-fly write • Up to 40% NTF CND Scratch • System-dependent TA Head degradation • Up to 30% • System-dependent, Grown defect personnel-dependent, Motor procedure-dependent, Mishandling Handling damage PCB etc. Observation: Tribology is responsible for many failure modes ! Andrei Khurshudov, 2007
  • 19. Tribology inside HDD Connectors FDB Motor Head-Disk Interface Ramp (friction and wear) Pivot Bearing Screws (wear and torque retention) There are multiple ways in which tribology impacts HDD reliability Andrei Khurshudov, 2007
  • 20. The Role of Tribology in HDD Reliability It is estimated that 15% to 35% of all HDD failures are linked to Tribology (25% on average) Improving tribological robustness enhances overall disk drive reliability Major known failure modes related to tribological issues: Scratch (on both head and media; with or w/out particles) Thermal erasure (disk) and head degradation New defects Weak write / read Crash Failure of some other moving parts Etc. Andrei Khurshudov, 2007
  • 21. Future Improvement Opportunities HDD reliability: Number of drives that will not fail between 2008 and 2012 per every 0.1% AFR improvement: ~ 3,000,000 Amount of stored information that will not be lost/impacted between 2008 and 2012 per every 0.1% AFR improvement: ~ 1,000,000 TB (or 1 EB) Tribology: Number of drives that will not fail between 2008 and 2012 due to Tribological problems per every 0.1% AFR improvement: ~ 750,000 Amount of stored information that will not be lost/impacted between 2008 and 2012 due to Tribological problems per every 0.1% AFR improvement: ~ 250,000 TB = 250 PB Andrei Khurshudov, 2007
  • 22. Is this worth the effort? Petabytes in use: The “American Memory” project is one of the largest digitized archives of U.S. history, with more than 7.5 million digital records from 100 collections of manuscripts, books, maps, films, sound recordings and photographs. The total size of the project is 0.008 Petabytes [Wired] As of November 2006, eBay had 2 Petabytes of data [Wikipedia] [Wikipedia Jefferson National Accelerator Facility has a 2 Petabyte storage farm used to collect data from experiments on the particle accelerator [Wikipedia] [Wikipedia RapidShare in 2007 had 3.5 Petabytes of hard-disk storage [Wikipedia] [Wikipedia The San Diego Supercomputer Center (SDSC) in the USA has a 1-Petabyte hard disk store and a 6-Petabyte robotic tape store [Wikipedia] [Wikipedia Microsoft stores on 900 servers a total of about 14 Petabytes. These are mostly imagery for Microsoft's digital model planet, Virtual Earth [Wikipedia] 15 Petabytes of data will be generated each year in particle physics experiments using CERN’s Large Hadron Collider, due to be launched in May 2008 [Wikipedia] [Wikipedia The total storage capacity needed for the above data is ~ 44 PB A failure rate reduction of 0.005% over the next 5 years is required to cover the above storage capacity needs Andrei Khurshudov, 2007
  • 23. Future Scenario Exponential growth of data over time (information avalanche) Lower cost of data storage per GB Many more disk drives required to accommodate all of the new data and backup Continually increasing reliability of disk drives Nevertheless, more total failures (in absolute terms) unless HDD reliability increases on a faster rate than the drive unit growth Andrei Khurshudov, 2007
  • 24. Conclusions Data storage capacity growth enables overall information growth Reliability of data storage devices is a key element in this growth Unreliability is extremely costly Even small improvements in reliability will have huge impact on the amount of information preserved in the future Tribology is, and will remain, a major enabler of the future information growth Relative contribution of Tribology to HDD unreliability is on the order of 25% Andrei Khurshudov, 2007
  • 25. References “The Innovator’s Dilemma” by Clayton M. Christensen Google: Failure Trends in a Large Disk Drive Population, E. Pinheiro, W.-D. Weber and L. Andr´e Barroso, FAST 2007 Wired: http://www.wired.com/science/discoveries/news/2002/10/55509 Wikipedia on Petabytes: http://en.wikipedia.org/wiki/Petabyte Consumer Reports National Research Center, 2006 Product Reliability Survey: http://www.squaretrade.com/htm/pop/lm_failureRates.html Proton rocket launcher: http://en.wikipedia.org/wiki/Proton_rocket HDD specifications: www.seagate.com Medical pacemaker’s reliability: http://www.medscape.com/viewarticle/536755 2001 Cost of Downtime Survey Results: http://www.datadepositbox.com/media/data- loss-statistics.asp BostonComputing.net: http://www.bostoncomputing.net/consultation/databackup/statistics IDC: IDC analyst Fred Broussard, PC Backup and Higher Prioritization for the Enterprise and Consumer, July 2002 Andrei Khurshudov, 2007
  • 26. Thank you! Andrei Khurshudov, 2007