![• Multiple machines for better scalability
Node 1 Node 2
[Read] Scale
11](https://image.slidesharecdn.com/asyncfosdem2020-200204205537/85/How-Safe-is-Asynchronous-Master-Master-Setup-23-320.jpg)
How Safe is Asynchronous Master-Master Setup?
- 1. How Safe is Asynchronous Master-Master February, 01, 2020 Sveta Smirnova
- 2. • MySQL Support engineer • Author of • MySQL Troubleshooting • JSON UDF functions • FILTER clause for MySQL • Speaker • Percona Live, OOW, Fosdem, DevConf, HighLoad... Sveta Smirnova 2
- 3. •What Happens with Two Active Nodes? •Why to Write to Multiple Nodes? •Galera, PXC, InnoDB Cluster •How to Setup Safe Asynchronous Master-Master Table of Contents 3
- 4. Master Slave ← Initiates Asynchronous Built-In Replication 4
- 5. Master Slave ← Initiates ← Requests a packet Asynchronous Built-In Replication 4
- 6. Master Sends the packet → Slave ← Initiates ← Requests a packet Asynchronous Built-In Replication 4
- 7. Master Sends the packet → Slave ← Initiates ← Requests a packet ... ? Asynchronous Built-In Replication 4
- 8. What Happens with Two Active Nodes?
- 9. Node1 INSERT SET id=42 Node2 INSERT SET id=42 INSERT 6
- 10. Node1 INSERT SET id=42 Sends the update → Node2 INSERT SET id=42 ← Sends the update INSERT 6
- 11. Node1 INSERT SET id=42 Sends the update → Receives the update → Node2 INSERT SET id=42 ← Sends the update ← Receives the update INSERT 6
- 12. Node1 INSERT SET id=42 Sends the update → Receives the update → Duplicate key error! Node2 INSERT SET id=42 ← Sends the update ← Receives the update Duplicate key error! INSERT 6
- 13. Node1 DELETE WHERE id=42 Node2 DELETE WHERE id=42 DELETE 7
- 14. Node1 DELETE WHERE id=42 Sends the update → Node2 DELETE WHERE id=42 ← Sends the update DELETE 7
- 15. Node1 DELETE WHERE id=42 Sends the update → Receives the update → Node2 DELETE WHERE id=42 ← Sends the update ← Receives the update DELETE 7
- 16. Node1 DELETE WHERE id=42 Sends the update → Receives the update → Key not found error! Node2 DELETE WHERE id=42 ← Sends the update ← Receives the update Key not found error! DELETE 7
- 17. Node1 UPDATE SET i=42 Node2 UPDATE SET i=25 UPDATE 8
- 18. Node1 UPDATE SET i=42 Sends the update → Node2 UPDATE SET i=25 ← Sends the update UPDATE 8
- 19. Node1 UPDATE SET i=42 Sends the update → Receives the update → Node2 UPDATE SET i=25 ← Sends the update ← Receives the update UPDATE 8
- 20. Node1 UPDATE SET i=42 Sends the update → Receives the update → i=25 Node2 UPDATE SET i=25 ← Sends the update ← Receives the update i=42 UPDATE 8
- 21. Why to Write to Multiple Nodes?
- 22. • Each node for its own purpose Application 1 Application 2 Data Distribution 10
- 23. • Multiple machines for better scalability Node 1 Node 2 [Read] Scale 11
- 24. • Nodes in different geographical regions Belgium Brazil GEO Distribution 12
- 25. • Waits when the main master deads • Then accepts writes Active Hot Standby Hot Standby 13
- 26. Galera, PXC, InnoDB Cluster
- 27. Node1 UPDATE ... Node2,3,... Fully Synchronous Cluster 15
- 28. Node1 UPDATE ... Sends the packet → Node2,3,... Fully Synchronous Cluster 15
- 29. Node1 UPDATE ... Sends the packet → Waits Node2,3,... ← Receives the packet Fully Synchronous Cluster 15
- 30. Node1 UPDATE ... Sends the packet → Waits Waits Node2,3,... ← Receives the packet Applies changes Fully Synchronous Cluster 15
- 31. Node1 UPDATE ... Sends the packet → Waits Waits Waits Node2,3,... ← Receives the packet Applies changes ← Confirms Fully Synchronous Cluster 15
- 32. Node1 UPDATE ... Sends the packet → Waits Waits Waits Receives confirmation → Node2,3,... ← Receives the packet Applies changes ← Confirms Fully Synchronous Cluster 15
- 33. Node1 UPDATE ... Sends the packet → Waits Waits Waits Receives confirmation → Ready for new update Node2,3,... ← Receives the packet Applies changes ← Confirms Fully Synchronous Cluster 15
- 34. • Speed of the slowest member Synchronization Penalty 16
- 35. • Speed of the slowest member • Writes or reads should wait Synchronization Penalty 16
- 36. • Speed of the slowest member • Writes or reads should wait • Tunable Synchronization Penalty 16
- 37. • Speed of the slowest member • Writes or reads should wait • Tunable • Galera • wsrep sync wait • InnoDB Cluster • group replication consistency Synchronization Penalty 16
- 38. • Synchronization takes too much time • Nodes often disconnect GEO Replication is Painful 17
- 39. How to Setup Safe Asynchronous Master-Master
- 40. • Business logic • Databases • Tables • Rows Divide and Rule 19
- 41. • Each server writes to its own set of tables Node 1 Node 2 D1 D2 D3 D4 D1 D2 D3 D4 Databases and Tables 20
- 42. • Agreement on which rows Node 1 Node 2 Row 1 Row 2 Row 3 Row 4 Row 1 Row 2 Row 3 Row 4 Rows 21
- 43. • Agreement on which rows • E.g. AUTO INCREMENT • auto increment increment=NUMBER OF SERVERS • auto increment offset = server id • Custom index • Create your own unique pattern for the each node Rows 21
- 44. • Hot Standby • Easy to setup and maintain • Separating writes by database/table • Comparatively easy to setup and maintain • Separating writes by row • Should be used very carefully Conclusion 22
- 45. MySQL User Reference Manual Why MySQL Replication Fails, and How to... MySQL High Availability More information 23