how to scale mongodb - dataops barcelona · how to scale mongodb ... • what is ha, how mongodb...

June-22-2018

How to Scale MongoDB

● Location: Skopje, Republic of Macedonia

● Education: MSc, Software Engineering

● Experience:

○ Lead Database Consultant (since 2016)

○ Database Consultant (2012 - 2016)

○ Web Developer, DBA (2007 - 2012)

● Certifications: C100DBA - MongoDB certified DBA (since 2016)

https://mk.linkedin.com/in/igorle @igorle

About me

© 2018 Pythian. Confidential

https://mk.linkedin.com/in/igorle

https://twitter.com/igorLE

Overview• What is HA, how MongoDB achieves HA

• Replica set, components, deployment topologies

• Scaling (Vertical vs Horizontal)

• What is a sharded cluster in MongoDB

• Cluster components - shards, config servers, mongos

• Shard keys and chunks

• Hashed vs range based sharding

• Choosing shard key

• QA


HA


What is HA?• High availability refers to systems that are durable and likely to operate

continuously without failure for a long time• single point of failure (SPOF) is a part of a system that, if it fails, will stop

the entire system from working


Application Database

SPOF? SPOF?

Database

Replica

What is HA?• HA on the Application vs Database layer

• Splitting Writes and Read operations


Application Primary

Application

…………...

ReplicaLoad

Bal

ance

r

Load

Bal

ance

r?

Writes

Reads

Database

Replica

What is HA?• Application node failure

• Database node failure

• Failover (elect new Primary node)


Application Primary

Application

…………...

ReplicaLoad

Bal

ance

r

Load

Bal

ance

r?

Writes

Reads

• Group of mongod processes that maintain

the same data set

• Redundancy and high availability

• Increased read capacity (scaling reads)

• Automatic failover

HA with MongoDB - replica sets


Replication concept1. Write operations go to the Primary node

2. All changes are recorded into operations log

3. Asynchronous replication to Secondary

4. Secondaries copy the Primary oplog

5. Secondary can use sync source Secondary


1.







2. oplog

1.







2. oplog

1.

3. 3.







2. oplog

1.

3. 3.

4. 4.





5. Secondary can use sync source Secondary*

*settings.chainingAllowed (true by default)


2. oplog

1.

3. 3.

4. 4.

5.

Configuration options• 50 members per replica set (7 voting members)

• Arbiter node

• Priority 0 node

• Hidden node

• Delayed node


• Does not hold copy of data

• Votes in elections

Arbiter node

hidden : true


Arbiter

Priority 0 nodePriority - floating point (i.e. decimal) number between 0 and 1000

• Cannot become primary, cannot trigger election

• Visible to application


Secondarypriority : 0

Hidden node• Not visible to application

• Never becomes primary

• Use cases○ reporting ○ backups

hidden : true


hidden: true priority:0

Secondaryhidden : true priority : 0

Delayed node• Must be priority 0 member

• Should be hidden member (not mandatory)

• Mainly used for backups

• Recovery in case of human error


SecondaryslaveDelay : 3600priority : 0hidden : true

Scaling


• Working set - amount of memory that a process requires in a given time

interval

• Time for scaling ?

Scaling


● Vertical scaling

○ Adding more power (CPU, RAM, DISK) to an existing machine

○ Might require downtime while scaling up

● Horizontal scaling

○ Adding more machines into your pool of resources

○ Partitioning the data on multiple machines

○ Parallelizing the processing

○ More complex to implement and maintain

Scaling


Scaling (vertical)


1TB

1TB

Scaling (horizontal) - Sharding


1TB

256GB 256GB 256GB 256GB

1TBShard 1 Shard 2 Shard 3 Shard 4

* https://www.amazon.com/dp/0596102356

Scaling costs


https://www.amazon.com/dp/0596102356

Scaling with MongoDB


Scaling (sharding) with MongoDB

Shard


Sharding with MongoDB

• Shard/Replica set

(subset of the sharded data)

• Config servers

(metadata and config settings)

• mongos

(query router, cluster interface)

mongos> sh.addShard("shardN")


1 2 N-1 N

Shards

• Contains subset of sharded data

• Replica set for redundancy and HA

• Primary shard

• Non sharded collections

• --shardsvr in config file (port 27018)


Config servers

• Stores the metadata for sharded cluster in

config database

• Authentication configuration information in

admin database

• Config servers as replica set only (>= 3.4)

• Holds balancer on Primary node (>= 3.4)

• --configsvr in config file (port 27019)


mongos• Caching metadata from config servers

• Routes queries to shards

• No persistent state

• Updates cache on metadata changes

• Holds balancer (mongodb <= 3.2)

• mongos version 3.4 cannot connect to

earlier mongod version


Sharding collection• Enable sharding on database "users"

sh.enableSharding("users")

• Shard collection historysh.shardCollection("users.history", { user_id : 1 } )

• Shard key - indexed key that exists in every document in the collection○ range based

sh.shardCollection("users.history", { user_id : 1 } )

○ hashed basedsh.shardCollection( "users.history", { user_id : "hashed" } )


Ranged sharding• Dividing data into contiguous ranges determined by the shard key values

• Documents with “close” shard key values are likely to be in the same

chunk or shard

• Query Isolation - more likely to target single shard for range queries


https://docs.mongodb.com/manual/reference/glossary/#term-chunk

Hashed sharding• Uses a hashed index of a single field as the shard key to partition data

• More even data distribution at the cost of reducing Query Isolation

• Applications do not need to compute hashes


Shard keys and chunksChoosing Shard key

• Large Shard Key Cardinality

• Low Shard Key Frequency

• Non-Monotonically changing shard keys (ranged or hashed sharding)

Chunks

• A contiguous range of shard key values within a particular shard

• Inclusive lower and exclusive upper range based on shard key

• Default size of 64MB ([1 .. 1024] MB)


Shard key cardinality


Shard key frequency


Monotonically changing shard key


Choosing Shard keyAnalyse your query patterns before making any decision (mtools - mloginfo)

● users○ Querying data by username (unique key?). Consider {username} as

shard key● user history

○ Querying data by user_id. Consider {user_id} as shard key○ Querying data by user_id, date_created. Consider {user_id,

date_created} as shard key● views

○ Write intensive workload. created_on (timestamp). Consider hashed sharding on {timestamp}


Shard keys limitations• Must be ascending indexed key or indexed compound keys that exists in

every document in the collection

• Cannot be multikey index, a text index or a geospatial index

• Cannot exceed 512 bytes

• Immutable key - can not be updated/changed

• Update operations that affect a single document must include the shard

key or the _id field

• No option for sharding if unique indexes on other fields exist

• No option for second unique index if the shard key is unique index


Summary• Replica set with odd number of voting members

• Hidden or Delayed member for dedicated functions (reporting, backups …)

• Dataset fits into single server, keep unsharded deployment

• Horizontal scaling - shards running as replica sets

• Shard keys are immutable with max size of 512 bytes

• Shard keys must exists in every document in the collection

• Ranged sharding may not distribute the data evenly

• Hashed sharding distributes the data randomly • Sharding adds complexity for maintenance (backups and upgrades)


Questions?


We’re Hiring!https://www.pythian.com/careers/


https://www.pythian.com/careers/