Kernel Engineer, 10genShaun Verch#MongoDBDaysIntroduction toSharding
Enable Sharding?> sh.enableSharding("production")Mon Jun 17 13:45:39.929 not connected to a mongosat src/mongo/shell/utils...
Agenda• Scaling Data• MongoDBs Approach• Architecture• Configuration• Mechanics
Scaling Data
Do you need to shard?• Hardware– Disks– Memory– Network– CPU• Application– Access patterns– Indexing– Schema design
Read/Write Throughput Exceeds I/O
Working Set Exceeds PhysicalMemoryWorking SetIndexesDataWorking SetIndexes Data
Vertical Scalability (Scale Up)
Horizontal Scalability (Scale Out)
Data Store Scalability• Custom Hardware– Oracle• Custom Software– Facebook + MySQL– Google
Data Store Scalability Today• MongoDB Auto-Sharding• A data store that is– Free– Publicly available– Open Source (https://...
MongoDBs Approachto Sharding
Partitioning• User defines shard key• Shard key defines range of data• Key space is like points on a line• Range is a segm...
Shards and Shard KeysShardShard keyrange
Data Distribution• Initially 1 chunk• Default max chunk size: 64mb• MongoDB automatically splits & migrates chunkswhen max...
Routing and Balancing• Queries routed to specificshards• MongoDB balances cluster• MongoDB migrates data tonew nodes
MongoDB Auto-Sharding• Minimal effort required– Same interface as single mongod• Two steps– Enable Sharding for a database...
Architecture
What is a Shard?• Shard is a node of the cluster• Shard can be a single mongod or a replica set
• Config Server– Stores cluster chunk ranges and locations– Can have only 1 or 3 (production must have 3)– Not a replica s...
Routing and Managing Data• Mongos– Acts as a router / balancer– No local data (persists to config database)– Can have 1 or...
Sharding infrastructure
Configuration
Example Cluster• Don’t use this setup in production!- Only one Config server (No FaultTolerance)- Shard not in a replica s...
Starting the Configuration Server• mongod --configsvr• Starts a configuration server on the default port (27019)
Start the mongos Router• mongos --configdb <hostname>:27019• For 3 configuration servers:mongos--configdb<host1>:<port1>,<...
Start the shard database• mongod --shardsvr• Starts a mongod with the default shard port (27018)• Shard is not yet connect...
Add the Shard• On mongos:- sh.addShard(„<host>:27018‟)• Adding a replica set:- sh.addShard(„<rsname>/<seedlist>‟)
Verify that the shard was added• db.runCommand({ listshards:1 }){ "shards" :[{"_id”:"shard0000”,"host”:”<hostname>:27018”}...
Enabling Sharding• Enable sharding on a databasesh.enableSharding(“<dbname>”)• Shard a collection with the given keysh.sha...
Tag Aware Sharding• Tag aware sharding allows you to control thedistribution of your data• Tag a range of shard keys– sh.a...
Mechanics
Partitioning• Remember its based on ranges
Chunk is a section of the entire range
Chunk splitting• Achunk is split once it exceeds the maximum size• There is no split point if all documents have the same ...
Balancing• Balancer is running on mongos• Once the difference in chunks between the most dense shardand the least dense sh...
Acquiring the Balancer Lock• The balancer on mongos takes out a “balancer lock”• To see the status of these locks:use conf...
Moving the chunk• The mongos sends a moveChunk command to source shard• The source shard then notifies destination shard• ...
Committing Migration• When complete, destination shard updates config server- Provides new locationsof the chunks
Cleanup• Source shard deletes moved data- Must wait for open cursors to either close or time out- NoTimeoutcursors may pre...
Routing Requests
Cluster Request Routing• Targeted Queries• Scatter Gather Queries• Scatter Gather Queries with Sort
Cluster Request Routing: TargetedQuery
Routable request received
Request routed to appropriate shard
Shard returns results
Mongos returns results to client
Cluster Request Routing: Non-TargetedQuery
Non-Targeted Request Received
Request sent to all shards
Shards return results to mongos
Mongos returns results to client
Cluster Request Routing: Non-TargetedQuery with Sort
Non-Targeted request with sortreceived
Request sent to all shards
Query and sort performed locally
Shards return results to mongos
Mongos merges sorted results
Mongos returns results to client
Shard Key
What is a Shard Key• Shard key is used to partition your collection• Shard key must exist in every document• Shard key is ...
Shard Key Considerations• Cardinality• Write Distribution• Query Isolation• Reliability• Index Locality
Conclusion
Read/Write Throughput Exceeds I/O
Working Set Exceeds PhysicalMemoryWorking SetIndexesDataWorking SetIndexes Data
Sharding Enables Scale• MongoDB‟sAuto-Sharding– Easy to Configure– Consistent Interface– Free and Open Source
• What‟s next?– Advanced Sharding Features in MongoDB 2.4 - JeremyMikola– MongoDB User Groups• Resourceshttps://education....
Kernel Engineer, 10genShaun Verch#MongoDBDaysThank You
Upcoming SlideShare
Loading in...5
×

Basic Sharding in MongoDB presented by Shaun Verch

1,992

Published on

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

No Downloads
Views
Total Views
1,992
On Slideshare
0
From Embeds
0
Number of Embeds
5
Actions
Shares
0
Downloads
67
Comments
0
Likes
4
Embeds 0
No embeds

No notes for slide
  • Ops will be most interested in ConfigurationDev will be most interested in Mechanics
  • Ops will be most interested in ConfigurationDev will be most interested in Mechanics
  • The goal here is to talk about data growth in general as setup for the problem that sharding solves.Take this slide and toss it out the window for the combined sharding talk. It’s a nice intro, but it’s more effective to spend time talking about tag aware and shard keys if you’re trying to go beyond the intro talk.
  • The goal here is to talk about data growth in general as setup for the problem that sharding solves.Take this slide and toss it out the window for the combined sharding talk. It’s a nice intro, but it’s more effective to spend time talking about tag aware and shard keys if you’re trying to go beyond the intro talk.
  • Working Set = most frequently accessed Indexes + Data
  • From mainframes, to RAC Oracle servers... People solved problems by adding more resources to a single machine.
  • Large scale operation can be combined with high performance on commodity hardware through horizontal scalingBuild - Document oriented database maps perfectly to object oriented languagesScale - MongoDB presents clear path to scalability that isn&apos;t ops intensive - Provides same interface for sharded cluster as single instance
  • In the past, you’ve had two ways to achieve datastore scalability:Lots of money to purchase custom hardwareLots of money to develop custom software
  • Today, MongoDB helps you achieve scalability without the need for custom software of specialized hardware.
  • Each “shard” in an encyclopedia is approximately the same size, ~1000 pages.
  • MongoDB 2.2 and later only need &lt;host&gt; and &lt;port&gt; for one member of the replica set
  • This can be skipped for the intro talk, but might be good to include if you’re doing the combined sharding talk. Totally optional, you don’t really have enough time to do this topic justice but it might be worth a mention.
  • Quick review from earlier.
  • 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
  • Balancer lock actually held on config server.
  • Moved chunk on shard2 should be gray
  • How do the other mongoses know that their configuration is out of date? When does the chunk version on the shard itself get updated?
  • The mongos does not have to load the whole set into memory since each shard sorts locally. The mongos can just getMore from the shards as needed and incrementally return the results to the client.
  • Isolating queries (to a few shards)Scatter -- gather ( high latency but not bad )hash keys
  • Cardinality – Can your data be broken down enough?Query Isolation - query targeting to a specific shardReliability – shard outagesA good shard key can:Optimize routingMinimize (unnecessary) trafficAllow best scaling
  • This section can be skipped for the intro talk, but might be good to include if you’re doing the combined sharding talk.
  • Reliability = if we lost an entire shard what would be the effect on the system?For most frequently used/largest indexes, how are they partitioned across shards?
  • Reliability = if we lost an entire shard what would be the effect on the system?For most frequently used/largest indexes, how are they partitioned across shards?
  • Reliability = if we lost an entire shard what would be the effect on the system?For most frequently used/largest indexes, how are they partitioned across shards?
  • Reliability = if we lost an entire shard what would be the effect on the system?For most frequently used/largest indexes, how are they partitioned across shards?
  • Compound shard key, utilize already existing (used) index, partition chunk for &quot;power&quot; users.For those users more than one shard *may* need to be queried in a targeted query.Reliability = if we lost an entire shard what would be the effect on the system?For most frequently used/largest indexes, how are they partitioned across shards?
  • Working Set = most frequently accessed Indexes + Data
  • Basic Sharding in MongoDB presented by Shaun Verch

    1. 1. Kernel Engineer, 10genShaun Verch#MongoDBDaysIntroduction toSharding
    2. 2. Enable Sharding?> sh.enableSharding("production")Mon Jun 17 13:45:39.929 not connected to a mongosat src/mongo/shell/utils_sh.js:6>
    3. 3. Agenda• Scaling Data• MongoDBs Approach• Architecture• Configuration• Mechanics
    4. 4. Scaling Data
    5. 5. Do you need to shard?• Hardware– Disks– Memory– Network– CPU• Application– Access patterns– Indexing– Schema design
    6. 6. Read/Write Throughput Exceeds I/O
    7. 7. Working Set Exceeds PhysicalMemoryWorking SetIndexesDataWorking SetIndexes Data
    8. 8. Vertical Scalability (Scale Up)
    9. 9. Horizontal Scalability (Scale Out)
    10. 10. Data Store Scalability• Custom Hardware– Oracle• Custom Software– Facebook + MySQL– Google
    11. 11. Data Store Scalability Today• MongoDB Auto-Sharding• A data store that is– Free– Publicly available– Open Source (https://github.com/mongodb/mongo)– Horizontally scalable– Application independent
    12. 12. MongoDBs Approachto Sharding
    13. 13. Partitioning• User defines shard key• Shard key defines range of data• Key space is like points on a line• Range is a segment of that line
    14. 14. Shards and Shard KeysShardShard keyrange
    15. 15. Data Distribution• Initially 1 chunk• Default max chunk size: 64mb• MongoDB automatically splits & migrates chunkswhen max reached
    16. 16. Routing and Balancing• Queries routed to specificshards• MongoDB balances cluster• MongoDB migrates data tonew nodes
    17. 17. MongoDB Auto-Sharding• Minimal effort required– Same interface as single mongod• Two steps– Enable Sharding for a database– Shard collection within database
    18. 18. Architecture
    19. 19. What is a Shard?• Shard is a node of the cluster• Shard can be a single mongod or a replica set
    20. 20. • Config Server– Stores cluster chunk ranges and locations– Can have only 1 or 3 (production must have 3)– Not a replica setMeta Data Storage
    21. 21. Routing and Managing Data• Mongos– Acts as a router / balancer– No local data (persists to config database)– Can have 1 or many
    22. 22. Sharding infrastructure
    23. 23. Configuration
    24. 24. Example Cluster• Don’t use this setup in production!- Only one Config server (No FaultTolerance)- Shard not in a replica set (LowAvailability)- Only one mongos and shard (No Performance Improvement)- Useful for developmentor demonstrating configuration mechanics
    25. 25. Starting the Configuration Server• mongod --configsvr• Starts a configuration server on the default port (27019)
    26. 26. Start the mongos Router• mongos --configdb <hostname>:27019• For 3 configuration servers:mongos--configdb<host1>:<port1>,<host2>:<port2>,<host3>:<port3>• This is always how to start a new mongos, even if the clusteris already running
    27. 27. 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
    28. 28. Add the Shard• On mongos:- sh.addShard(„<host>:27018‟)• Adding a replica set:- sh.addShard(„<rsname>/<seedlist>‟)
    29. 29. Verify that the shard was added• db.runCommand({ listshards:1 }){ "shards" :[{"_id”:"shard0000”,"host”:”<hostname>:27018”} ],"ok" : 1}
    30. 30. Enabling Sharding• Enable sharding on a databasesh.enableSharding(“<dbname>”)• Shard a collection with the given keysh.shardCollection(“<dbname>.people”,{“country”:1})• Use a compound shard key to prevent duplicatessh.shardCollection(“<dbname>.cars”,{“year”:1,”uniqueid”:1})
    31. 31. Tag Aware Sharding• Tag aware sharding allows you to control thedistribution of your data• Tag a range of shard keys– sh.addTagRange(<collection>,<min>,<max>,<tag>)• Tag a shard– sh.addShardTag(<shard>,<tag>)
    32. 32. Mechanics
    33. 33. Partitioning• Remember its based on ranges
    34. 34. Chunk is a section of the entire range
    35. 35. Chunk splitting• Achunk is split once it exceeds the maximum size• There is no split point if all documents have the same shardkey• Chunk split is a logical operation (no data is moved)
    36. 36. Balancing• Balancer is running on mongos• Once the difference in chunks between the most dense shardand the least dense shard is above the migration threshold, abalancing round starts
    37. 37. Acquiring the Balancer Lock• The balancer on mongos takes out a “balancer lock”• To see the status of these locks:use configdb.locks.find({_id: “balancer”})
    38. 38. Moving the chunk• The mongos sends a moveChunk command to source shard• The source shard then notifies destination shard• Destination shard starts pulling documents from source shard
    39. 39. Committing Migration• When complete, destination shard updates config server- Provides new locationsof the chunks
    40. 40. Cleanup• Source shard deletes moved data- Must wait for open cursors to either close or time out- NoTimeoutcursors may preventthe release of the lock• The mongos releases the balancer lock after old chunks aredeleted
    41. 41. Routing Requests
    42. 42. Cluster Request Routing• Targeted Queries• Scatter Gather Queries• Scatter Gather Queries with Sort
    43. 43. Cluster Request Routing: TargetedQuery
    44. 44. Routable request received
    45. 45. Request routed to appropriate shard
    46. 46. Shard returns results
    47. 47. Mongos returns results to client
    48. 48. Cluster Request Routing: Non-TargetedQuery
    49. 49. Non-Targeted Request Received
    50. 50. Request sent to all shards
    51. 51. Shards return results to mongos
    52. 52. Mongos returns results to client
    53. 53. Cluster Request Routing: Non-TargetedQuery with Sort
    54. 54. Non-Targeted request with sortreceived
    55. 55. Request sent to all shards
    56. 56. Query and sort performed locally
    57. 57. Shards return results to mongos
    58. 58. Mongos merges sorted results
    59. 59. Mongos returns results to client
    60. 60. Shard Key
    61. 61. What is a Shard Key• Shard key is used to partition your collection• Shard key must exist in every document• Shard key is immutable• Shard key values are immutable• Shard key must be indexed• Shard key is used to route requests to shards
    62. 62. Shard Key Considerations• Cardinality• Write Distribution• Query Isolation• Reliability• Index Locality
    63. 63. Conclusion
    64. 64. Read/Write Throughput Exceeds I/O
    65. 65. Working Set Exceeds PhysicalMemoryWorking SetIndexesDataWorking SetIndexes Data
    66. 66. Sharding Enables Scale• MongoDB‟sAuto-Sharding– Easy to Configure– Consistent Interface– Free and Open Source
    67. 67. • What‟s next?– Advanced Sharding Features in MongoDB 2.4 - JeremyMikola– MongoDB User Groups• Resourceshttps://education.10gen.com/http://www.10gen.com/presentations
    68. 68. Kernel Engineer, 10genShaun Verch#MongoDBDaysThank You
    1. A particular slide catching your eye?

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

    ×