Seagate Kinetic Open Storage Platform
James Hughes
…and many others
Storage is a Price Elastic Market
http://en.wikipedia.org/wiki/Alfred_Marshall

Price elasticity of demand
• Alfred Marsha...
Areal Density Growth

• 

100000

Perpendicular expected to extend
to 0.5-1 Tb/in2

10000

Additional innovations required...
Cloud Computing will increase this trend
•Jevons Paradox
• Cloud Computing increases the efficiency of computing....

Seag...
Cloud Computing will increase this trend
•Jevons Paradox
• Cloud Computing increases the efficiency of computing....

Impr...
Technology TrendsNDA
Seagate Confidential: Subject to
No. 77103, effective Jan. 18, 2009,
and all applicable supplements

...
Shingled Disks
•Write head larger than read
head
• Turns Disk into a sequentially
written media
Seagate Confidential: Subj...
Log Structured Storage
How much is erased on a reposition?
• Tape - the remainder of the tape
• Shingled disk - the remain...
Single System Performance Trend
Leading to disaggregation of servers

http://web.eecs.umich.edu/~twenisch/papers/isca09-di...
Scaling Storage
Distributed Hash Table
• Key/Value Store

RAM
Flash
Disk

Memcached
FAWN
Riak

http://en.wikipedia.org/wik...
Metadata and Metadata Servers are Evil
Required by traditional file systems (POSIX) to translate
names to sectors
• Hard t...
Cumulative operations ordered by length
100%

operations
92% of the operations

80%

60%

40%

32KB

Cumulative percentage...
512KB
Length

Map of Operations

0

0
1

3

Tim
e

(m

2
inu
tes

)

3
4

0

1
oc
L

2
B)
(T
on
ati

"13
Seagate Kinetic
Seagate Confidential: Subject to NDA
Open Storage Platform
No. 77103, effective Jan. 18, 2009,
and all app...
Seagate Kinetic Open Storage Platform
Dis-intermediates applications to drive
–Goes around file systems, volume managers, ...
"16
"16
D

A
S

"17
App

Proprietary
to System Vendor

• Application
• Clustering
• Management

App
D

LibKinetic

C++, Java, Python, Erlang, ...
App

App

App

App

App

Proprietary
to System Vendor

• Application
• Clustering
• Management

App

D

LibKinetic

LibKin...
App

App

App

App

App

D

LibKinetic

LibKinetic

LibKinetic
GPL
Standard

ProtoBuf
TCP/IP/GbE

• Interconnect

D

A

D
...
System Hardware
Typical JBOD architecture
• Does not require a server, just JBODs to the ToR Switch
• 10 JBODS × 60 drives...
Features
Provides RPC to Key/Value database
• Data is pre-indexed
• Compression and other value is easy and transparent

P...
Kinetic Systems
Clustering (performance, reliability, management)
Compatibility with large scale applications (S3, etc.)
C...
Lower TCO
Elimination of server layers
Less Human requirements
Reduced mistakes
Disaggregate storage from
servers
Power ma...
Lower TCO
Elimination of server layers
Less Human requirements
Reduced mistakes
Disaggregate storage from
servers
Power ma...
Goals of API
Data movement
• Get/put/delete/getnext/getprevious
• Versioned (== for success), options

Range operations
Mu...
Management (System Vendor)
Configures the drive
• Network
• authorized clients

Monitors
• Health
• Statistics
• Logs

Ini...
Data formats
Key Structure
• Variable number of octets (0-4KB)

Data Structure (Serialized to a byte stream)
• KeyOf
• Ver...
Performance Metrics
Same normal performance expectations
•
•
•
•

Sequential Write
Random Write
Sequential Read
Sequential...
Demo Time!

Seagate Confidential: Subject to NDA
No. 77103, effective Jan. 18, 2009,
and all applicable supplements

"26
Performance Results

MB/s

Puts/s

120

1000

90

750

60

500

30

250

0

0
0

2

4

6

8

1MB values put rate (MB/s)

0...
Conclusion
Deliver more value to Seagate, Partners and Customers
• Dis-intermediates cloud applications to drive
• Enable ...
Upcoming SlideShare
Loading in …5
×

Kinetic basho public

1,177 views

Published on

Published in: Technology
0 Comments
3 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
1,177
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
16
Comments
0
Likes
3
Embeds 0
No embeds

No notes for slide

Kinetic basho public

  1. 1. Seagate Kinetic Open Storage Platform James Hughes …and many others
  2. 2. Storage is a Price Elastic Market http://en.wikipedia.org/wiki/Alfred_Marshall Price elasticity of demand • Alfred Marshall (1890) As the price of Storage approaches $0 • Demands for storage will approach infinity If the price of a Cisco router approaches $0 • Demands for routers will not approach infinity - Storage is different "3
  3. 3. Areal Density Growth •  100000 Perpendicular expected to extend to 0.5-1 Tb/in2 10000 Additional innovations required at that point •  heat-assisted recording •  bit patterned media recording Single particle superparamagnetic limit (estimated) 40% HAMR +BPM 1000 HAMR Perpendicular Writing & GMR 100 Seagate Confidential: Subject to NDA 100% 10 Charap’s limiteffective Jan. 18, 2009, No. 77103, (broken) and all applicable supplements GMR reading •  Inductive Writing/ 1 0.1 29% •  Inductive Writing/ MR reading •  Inductive Writing & Reading 19 89 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 20 07 20 09 20 11 20 13 20 15 20 17 20 19 •  Late 1990s – super paramagnetic limit demonstrated through modeling gigabit / in2 •  •  Areal Density CAGR 40% •  Transfer Rate CAGR 20% year "4
  4. 4. Cloud Computing will increase this trend •Jevons Paradox • Cloud Computing increases the efficiency of computing.... Seagate Confidential: Subject to NDA No. 77103, effective Jan. 18, 2009, and all applicable supplements http://en.wikipedia.org/wiki/Jevons_paradox "5
  5. 5. Cloud Computing will increase this trend •Jevons Paradox • Cloud Computing increases the efficiency of computing.... Improved technology doubles Seagate Confidential: Subject to NDA the amount of Information No. 77103, effective Jan. produced 18, with a given amount of Storage 2009, ! and all applicable supplements Demand for Storage rises http://en.wikipedia.org/wiki/Jevons_paradox "5
  6. 6. Technology TrendsNDA Seagate Confidential: Subject to No. 77103, effective Jan. 18, 2009, and all applicable supplements "6
  7. 7. Shingled Disks •Write head larger than read head • Turns Disk into a sequentially written media Seagate Confidential: Subject to NDA No. 77103, effective Jan. 18, 2009, and all applicable supplements •All updates to data and metadata are written sequentially to a continuous stream, called a log •Disk API of sectors is no longer “natural” http://www.ssrc.ucsc.edu/Papers/amer-ieeetm11.pdf "7
  8. 8. Log Structured Storage How much is erased on a reposition? • Tape - the remainder of the tape • Shingled disk - the remainder of the track group • Flash - the entire page All persistent Storage systems do/will implement log structure • e.g. “NoSQL Database of sectors” Does it make sense to layer a database on top of a database? • Could we use the log structure of the media to provide a more natural storage systems, not mimicking an antique paradigm? "8
  9. 9. Single System Performance Trend Leading to disaggregation of servers http://web.eecs.umich.edu/~twenisch/papers/isca09-disaggregate.pdf "9
  10. 10. Scaling Storage Distributed Hash Table • Key/Value Store RAM Flash Disk Memcached FAWN Riak http://en.wikipedia.org/wiki/Distributed_hash_table "10
  11. 11. Metadata and Metadata Servers are Evil Required by traditional file systems (POSIX) to translate names to sectors • Hard to scale, heavy HA requirements, expensive Can we use a name as a key? • Place the data into a scaled key value store? • Eliminate costly metadata servers? "11
  12. 12. Cumulative operations ordered by length 100% operations 92% of the operations 80% 60% 40% 32KB Cumulative percentage data 20% 0.5% of the data 0% 1.00 10.00 100.00 1000.00 10000.00 100000.00 Length (KB) "12
  13. 13. 512KB Length Map of Operations 0 0 1 3 Tim e (m 2 inu tes ) 3 4 0 1 oc L 2 B) (T on ati "13
  14. 14. Seagate Kinetic Seagate Confidential: Subject to NDA Open Storage Platform No. 77103, effective Jan. 18, 2009, and all applicable supplements "14
  15. 15. Seagate Kinetic Open Storage Platform Dis-intermediates applications to drive –Goes around file systems, volume managers, drivers Enable ecosystem of value added software –Partners (like Basho) can create their own system value Lower TCO –Eliminates complexity "15
  16. 16. "16
  17. 17. "16
  18. 18. D A S "17
  19. 19. App Proprietary to System Vendor • Application • Clustering • Management App D LibKinetic C++, Java, Python, Erlang, DIY GPL Standard ProtoBuf TCP/IP/GbE • Interconnect Proprietary to Seagate A • Storage S "17
  20. 20. App App App App App Proprietary to System Vendor • Application • Clustering • Management App D LibKinetic LibKinetic LibKinetic GPL Standard ProtoBuf TCP/IP/GbE • Interconnect Proprietary to Seagate A • Storage S "17
  21. 21. App App App App App D LibKinetic LibKinetic LibKinetic GPL Standard ProtoBuf TCP/IP/GbE • Interconnect D A D A S D A S D A A S S D A S D A S D A S A S S "17 Proprietary to Seagate D • Storage App Proprietary to System Vendor • Application • Clustering • Management
  22. 22. System Hardware Typical JBOD architecture • Does not require a server, just JBODs to the ToR Switch • 10 JBODS × 60 drives × 4TB = 2.4PB/Rack "18
  23. 23. Features Provides RPC to Key/Value database • Data is pre-indexed • Compression and other value is easy and transparent P2P (Drive to Drive) copy of key ranges Communicate using existing Data Center Plumbing (TCP/IP) Multiple masters - Data sharing between machines Configurable caching per command • Async, Sync, Flush Local space management "19
  24. 24. Kinetic Systems Clustering (performance, reliability, management) Compatibility with large scale applications (S3, etc.) Centralized Management • Reliability, availability, durability "20
  25. 25. Lower TCO Elimination of server layers Less Human requirements Reduced mistakes Disaggregate storage from servers Power management "21
  26. 26. Lower TCO Elimination of server layers Less Human requirements Reduced mistakes Disaggregate storage from servers Power management "21
  27. 27. Goals of API Data movement • Get/put/delete/getnext/getprevious • Versioned (== for success), options Range operations Multiple masters • Authentication/Integrity/Authorization Cluster-able • Simple cluster configuration version enforcement 3rd party copy Management "22
  28. 28. Management (System Vendor) Configures the drive • Network • authorized clients Monitors • Health • Statistics • Logs Initiates recovery • Change cluster version • 3rdPartyCopy "23
  29. 29. Data formats Key Structure • Variable number of octets (0-4KB) Data Structure (Serialized to a byte stream) • KeyOf • Version • E2E Data Integrity –Algorithm name • Data Variable length (0-xMB) "24
  30. 30. Performance Metrics Same normal performance expectations • • • • Sequential Write Random Write Sequential Read Sequential Write Iometer for key/value "25
  31. 31. Demo Time! Seagate Confidential: Subject to NDA No. 77103, effective Jan. 18, 2009, and all applicable supplements "26
  32. 32. Performance Results MB/s Puts/s 120 1000 90 750 60 500 30 250 0 0 0 2 4 6 8 1MB values put rate (MB/s) 0 2 4 6 1KB values put rate "27 8
  33. 33. Conclusion Deliver more value to Seagate, Partners and Customers • Dis-intermediates cloud applications to drive • Enable innovation in hardware and software ecosystem • Lower TCO OpenSource Software –Basho Riak, Swift, HDFS More information • http://seagate.com/www/kinetic • https://developers.seagate.com • http://guthub.com/Seagate "28

×