How Do I Configure the Shard Key for a MongoDB Sharded Cluster?

MongoDB shards data at the collection level, distributing the collection data using shard keys.

You choose the shard key when sharding a collection. Each record contains a shard key, and the shard key is either an indexed field or indexed compound fields. MongoDB database distributes data in different chunks according to the shard key, and distributes chunks evenly among the shards. To divide data chunks by shard key, MongoDB database uses two sharding methods: range-based sharding and hashed sharding.

**Table 1** Shard key classification
Shard Key Type	Description	Application Scenario
Range-based sharding	Ranged-based sharding involves dividing data into contiguous ranges determined by the shard key values. Range-based sharding is the default sharding methodology if no other options are specified. This allows for efficient queries where reads target documents within a contiguous range. The distribution route determines which data chunk stores the data required and forwards the request to the corresponding shard.	It is recommended when the shard key has high cardinality with low frequency, and the shard key value does not change monotonically.
Hashed sharding	Hashed sharding uses a hashed index to partition data across your shared cluster and to create chunks. Hashed sharding provides more even data distribution across the sharded cluster Hash values enable data to be randomly distributed in each chunk, and therefore are randomly distributed in different shards.	If the shard key values that have a high cardinality or change monotonically, or there are large number of different values, hashed sharding is an ideal option.

Once you shard a collection, the shard key and the shard key values are immutable. If you need to modify the shard key of a document, you must delete the document. Then modify the shard key and insert the document again.

The shard key does not support array indexes, text indexes, geographical indexes, and spatial indexes.

Range-based Sharding

Run the following command to enable database sharding:

sh.enableSharding(database)

database indicates the database for which the sharded collection is enabled.
Configure the collection's shard key.

sh.shardCollection(namespace, key)
- namespace consists of a string <database>.<collections> specifying the full namespace of the target collection.
- key indicates the index for the shard key.
- If the collection is empty, skip this step because the index on the shard key can be created automatically.
  sh.shardCollection()
- If the collection is not empty, create an index key. Then, run the following command to set the shard key:
  sh.shardCollection()

Hashed Sharding

Run the following command to enable database sharding:

sh.enableSharding(database)

database indicates the database for which the sharded collection is enabled.
Set hashed shard keys.

sh.shardCollection("<database>.<collection>", { <shard key> : "hashed" } , false, {numInitialChunks: Number of preconfigured chunks})

The value of numInitialChunks is calculated as follows: db.collection.stats().size / 10*1024*1024*1024.

If the collection contains data, run the following command to create a hashed index for the hashed key:

db.collection.createIndex()

Run the following command to create a hashed shard key:

sh.shardCollection()