Published on

For the first time this year, 10gen will be offering a track completely dedicated to Operations at MongoSV, 10gen's annual MongoDB user conference on December 4. Learn more at

  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • Ops will be most interested in ConfigurationDev will be most interested in Mechanics
  • Build - Document oriented database maps perfectly to object oriented languagesScale - MongoDB presents clear path to scalability that isn't ops intensive - Provides same interface for sharded cluster as single instance
  • In 2005, two ways to achieve datastore scalability:Lots of money to purchase custom hardwareLots of money to develop custom software
  • Indexes should be contained in working set.
  • The solution to the preceding problemsAdd graphic??
  • Add arrows for or mention that there is communication between shards (migrations)
  • Once chunk size is reached, mongos asks mongod to split a chunk + internal function called splitVector()mongod counts number of documents on each side of split + based on avg. document size `db.stats()`Chunk split is a **logical** operation (no data has moved)Max on first chunk should be 14
  • Moved chunk on shard2 should be gray
  • Use this slide to talk about clients that have had this problem and worked around it by sharding.
  • Use this slide to talk about clients that have had this problem and worked around it by sharding.
  • Sharding

    1. 1. #ConferenceHashtagShardingSpeaker NameJob Title, 10gen
    2. 2. Agenda• Scaling data• Why shard• Architecture• Configuration• Mechanics• Solutions
    3. 3. The story of scalingdata
    4. 4. Visual representation of vertical scalingAltaVista, 1995: Vertical Scalability
    5. 5. Visual representation of horizontal scalingGoogle, ~2000: Horizontal Scalability
    6. 6. Data Store Scalability in 2005• Custom Hardware – Oracle• Custom Software – Facebook + MySQL
    7. 7. Data Store Scalability Today• MongoDB auto-sharding available in 2009• A data store that is – Free – Publicly available – Open source – Horizontally scalable
    8. 8. Why shard?
    9. 9. Working Set Exceeds PhysicalMemory
    10. 10. Read/Write Throughput Exceeds I/O
    11. 11. MongoDB Auto-Sharding• Minimal effort required – Same interface as single mongod• Two steps – Enable Sharding for a database – Shard collection within database
    12. 12. Architecture
    13. 13. Architecture – Components• shard – Can be stand alone or replica set
    14. 14. Architecture – Components• Config Server – Stores cluster meta-data (chunk ranges and locations) – Can have only 1 or 3 (production must have 3) – Two phase commit (not a replica set)
    15. 15. Architecture – Components• Mongos – Acts as a router / balancer – No local data (persists to config database) – Can have 1 or many
    16. 16. Sharding infrastructure
    17. 17. Configuration
    18. 18. Example cluster setup• Don‟t use this setup in production! - Only one Config server (No Fault Tolerance) - Shard not in a replica set (Low Availability) - Only one Mongos and shard (No Performance Improvement)
    19. 19. Start the config server• “mongod --configsvr”• Starts a config server on the default port (27019)
    20. 20. Start the mongos router• “mongos --configdb <hostname>:27019”• For 3 config servers: “mongos --configdb <host1>:<port1>,<host2>:<port2>,<host3>:<port3>”• This is always how to start a new mongos, even if the cluster is already running
    21. 21. Start the shard database• “mongod --shardsvr”• Starts a mongod with the default shard port (27018)• Shard is not yet connected to the rest of the cluster• Shard may have already been running in production
    22. 22. Add the shard• On mongos: “sh.addShard(„<host>:27018‟)”• Adding a replica set: “sh.addShard(„<rsname>/<seedlist>‟)
    23. 23. Verify that the shard was added• db.runCommand({ listshards:1 })• { "shards" : [ { "_id”: "shard0000”, "host”: ”<hostname>:27018” } ], "ok" : 1 }
    24. 24. Enabling Sharding• Enable sharding on a database – sh.enableSharding(“records”)• Shard a collection with the given key – sh.shardCollection(“records.people”,{“country”:1})• Use a compound shard key to prevent duplicates – sh.shardCollection(“”,{“year”:1, ”uniqueid”:1})
    25. 25. Tag Aware Sharding• Tag aware sharding allows you to control the distribution of your data• Tag a range of shard keys – sh.addTagRange(<collection>,<min>,<max>,<tag>)• Tag a shard – sh.addShardTag(<shard>,<tag>)
    26. 26. Mechanics
    27. 27. Partitioning• User defined shard key• Range based partitioning• Initially 1 chunk• Default max chunk size: 64mb• Once max size is reached a split occurs
    28. 28. Partitioning
    29. 29. Chunk splitting• Balancer is running on mongos• Once the difference in chunks between the most dense shard and the least dense shard is above the migration threshold, a balancing round starts
    30. 30. Balancing• Balancer is running on mongos• Once the difference in chunks between the most dense shard and the least dense shard is above the migration threshold, a balancing round starts
    31. 31. Acquiring the Balancer Lock• The balancer on mongos takes out a “balancer lock”• To see the status of these locks: - use config - db.locks.find({ _id: “balancer” })
    32. 32. Moving the chunk• The mongos sends a “moveChunk” command to source shard• The source shard then notifies destination shard• The destination clears the chunk shard-key range• Destination shard starts pulling documents from source shard
    33. 33. Committing Migration• When complete, destination shard updates config server - Provides new locations of the chunks
    34. 34. Cleanup• Source shard deletes moved data - Must wait for open cursors to either close or time out - NoTimeout cursors may prevent the release of the lock• Mongos releases the balancer lock after old chunks are deleted
    35. 35. Cluster Request Routing• Targeted Queries• Scatter Gather Queries• Scatter Gather Queries with Sort
    36. 36. Cluster Request Routing: TargetedQuery
    37. 37. Routable request received
    38. 38. Request routed to appropriate shard
    39. 39. Shard returns results
    40. 40. Mongos returns results to client
    41. 41. Cluster Request Routing: Non-TargetedQuery
    42. 42. Non-Targeted Request Received
    43. 43. Request sent to all shards
    44. 44. Shards return results to mongos
    45. 45. Mongos returns results to client
    46. 46. Cluster Request Routing: Non-TargetedQuery with Sort
    47. 47. Non-Targeted request with sortreceived
    48. 48. Request sent to all shards
    49. 49. Query and sort performed locally
    50. 50. Shards return results to mongos
    51. 51. Mongos merges sorted results
    52. 52. Mongos returns results to client
    53. 53. Shard Key• Choose a field common to queries• Shard key is immutable• Shard key values are immutable• Shard key requires index on fields contained in key• Uniqueness of `_id` field is only guaranteed within individual shard• Shard key limited to 512 bytes in size
    54. 54. Shard Key Considerations• Cardinality• Write distribution• Query isolation• Data Distribution
    55. 55. Sharding enables scale
    56. 56. Working Set Exceeds PhysicalMemory
    57. 57. Read/Write throughput exceeds I/O Click tracking
    58. 58. #ConferenceHashTagThank YouSpeaker NameJob Title, 10gen