Your SlideShare is downloading. ×
0
Cell ArchitectureFor Large Scale Servicecharsyam@naver.com
Agenda•Basic Architecture•Query-Off Style Architecture•Cell Architecture
Basic ArchitectureCLIENT WEB DBRead/Write
Basic ArchitectureOverhead
OverheadWhatcausesOverhead
OverheadProfiling
ProfilingJavaScriptCSSCLIENT
ProfilingJavaScriptCSSCLIENTOptimization
ProfilingBusiness LogicWEB
ProfilingBusiness LogicWEBOptimizationIncrease WEB/AS
Second ArchitectureCLIENT WEB DBRead/WriteWEBWEB
Second ArchitectureEasy!!!
ProfilingStorage LayerDB
ProfilingStorage LayerDBI/O Consists ofRead/Write
ProfilingGenerallyREADs are morethan WRITEs
ProfilingDistribute READs
Profiling200 writes/s800 reads/sRead > Write
Simple DB ArchitectureMasterSlaveREPLICATION/FailOver
Query-Off ArchitectureWEB/ASMasterSlaveONLY WRITESlave SlaveOnly READREPLICATION
Query-Off Architecture200writes/s800reads/s200writes/s400reads/s200writes/s400reads/sRead/1 Server Read/2 Server
Query-Off ArchitectureIs performanceproportional tonumbers of Slaves?
Query-Off Architecture700writes/s50reads/s700writes/s50reads/s700writes/s50reads/s700writes/s50reads/s700writes/s50reads/s...
Query-Off ArchitectureZero-Sum Game
ProfilingDistribute Writes
How?
ClusteringandSharding
Clusteringseems to beIdeal
Clustering – Large oneWebWebWebDBDB DBDB
ClusteringDifficult
Sharding
ShardingWEB/ASMaster Master MasterUser A DataUser C DataUser D DataUser B DataUser X DataUser Z DataUser Y DataUser E Data...
Cell ArchitectureCell 0 Cell 1 Cell 2UserCell InfoServerID: CharSyamCellID: 1, Status: NormalGet/set
Cell ArchitectureA Cell is Full-SetCan serve Users
Cell Examples #1WEB/ASMaster SlaveWEB/AS WEB/AS
Cell Examples #2WEB/AS WEB/AS WEB/ASMasterSlave Slave SlaveREAD onlyWRITE only
Failure of Cell ArchitectureCell 0 Cell 1 Cell 2UserCell InfoServerID: CharSyamCellID: 1, Status: NormalCan’t Service
Failure of Cell ArchitectureCell 0 Cell 1 Cell 2UserCell InfoServerID: CharSyamCan’t responseGet/set
• Benefits–Easy to extend–Failure is limited to some Users in same cell.–Can deploy specific feature to some cell users.• ...
Q&A
Thank you
Upcoming SlideShare
Loading in...5
×

Cell architecture

809

Published on

1 Comment
4 Likes
Statistics
Notes
No Downloads
Views
Total Views
809
On Slideshare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
13
Comments
1
Likes
4
Embeds 0
No embeds

No notes for slide

Transcript of "Cell architecture"

  1. 1. Cell ArchitectureFor Large Scale Servicecharsyam@naver.com
  2. 2. Agenda•Basic Architecture•Query-Off Style Architecture•Cell Architecture
  3. 3. Basic ArchitectureCLIENT WEB DBRead/Write
  4. 4. Basic ArchitectureOverhead
  5. 5. OverheadWhatcausesOverhead
  6. 6. OverheadProfiling
  7. 7. ProfilingJavaScriptCSSCLIENT
  8. 8. ProfilingJavaScriptCSSCLIENTOptimization
  9. 9. ProfilingBusiness LogicWEB
  10. 10. ProfilingBusiness LogicWEBOptimizationIncrease WEB/AS
  11. 11. Second ArchitectureCLIENT WEB DBRead/WriteWEBWEB
  12. 12. Second ArchitectureEasy!!!
  13. 13. ProfilingStorage LayerDB
  14. 14. ProfilingStorage LayerDBI/O Consists ofRead/Write
  15. 15. ProfilingGenerallyREADs are morethan WRITEs
  16. 16. ProfilingDistribute READs
  17. 17. Profiling200 writes/s800 reads/sRead > Write
  18. 18. Simple DB ArchitectureMasterSlaveREPLICATION/FailOver
  19. 19. Query-Off ArchitectureWEB/ASMasterSlaveONLY WRITESlave SlaveOnly READREPLICATION
  20. 20. Query-Off Architecture200writes/s800reads/s200writes/s400reads/s200writes/s400reads/sRead/1 Server Read/2 Server
  21. 21. Query-Off ArchitectureIs performanceproportional tonumbers of Slaves?
  22. 22. Query-Off Architecture700writes/s50reads/s700writes/s50reads/s700writes/s50reads/s700writes/s50reads/s700writes/s50reads/sWrite Heavy Situation
  23. 23. Query-Off ArchitectureZero-Sum Game
  24. 24. ProfilingDistribute Writes
  25. 25. How?
  26. 26. ClusteringandSharding
  27. 27. Clusteringseems to beIdeal
  28. 28. Clustering – Large oneWebWebWebDBDB DBDB
  29. 29. ClusteringDifficult
  30. 30. Sharding
  31. 31. ShardingWEB/ASMaster Master MasterUser A DataUser C DataUser D DataUser B DataUser X DataUser Z DataUser Y DataUser E DataUser F Data
  32. 32. Cell ArchitectureCell 0 Cell 1 Cell 2UserCell InfoServerID: CharSyamCellID: 1, Status: NormalGet/set
  33. 33. Cell ArchitectureA Cell is Full-SetCan serve Users
  34. 34. Cell Examples #1WEB/ASMaster SlaveWEB/AS WEB/AS
  35. 35. Cell Examples #2WEB/AS WEB/AS WEB/ASMasterSlave Slave SlaveREAD onlyWRITE only
  36. 36. Failure of Cell ArchitectureCell 0 Cell 1 Cell 2UserCell InfoServerID: CharSyamCellID: 1, Status: NormalCan’t Service
  37. 37. Failure of Cell ArchitectureCell 0 Cell 1 Cell 2UserCell InfoServerID: CharSyamCan’t responseGet/set
  38. 38. • Benefits–Easy to extend–Failure is limited to some Users in same cell.–Can deploy specific feature to some cell users.• Liabilities–To need more servers.• To build full-setCell Architecture
  39. 39. Q&A
  40. 40. Thank you
  1. A particular slide catching your eye?

    Clipping is a handy way to collect important slides you want to go back to later.

×