MongoDB -Sharding basics Horizontal Scaling Technique 

MongoDB sharding horizontally splits your data across multiple shards (each typically a replica set), so a 4‑shard cluster spreads one logical collection’s data over 4 independent data-bearing groups of nodes.mongodb+1​

Sharding basics

• Sharding is MongoDB’s horizontal scaling method: a single logical dataset is partitioned and stored on multiple servers called shards.geeksforgeeks+1​
  
  

• A sharded cluster has three main components:
  
  

  • Shards: store subsets of your data (each shard is a replica set).mongodb+1​
    
    

  • Config servers (3‑member CSRS): store metadata like chunk ranges and shard locations.geeksforgeeks+1​
    
    

  • Mongos routers: your applications connect here; mongos routes reads/writes to the right shard.mongodb+1​
    
    

Shard key, chunks, and balancer

• When you shard a collection, you must choose a shard key (one or more fields) that determines how documents are partitioned.devopedia+1​
  
  

• MongoDB divides the shard key space into ranges called chunks; each chunk covers an inclusive lower and exclusive upper bound of shard key values.bmc+1​
  
  

• The balancer process runs in the background, moving chunks between shards so each shard holds roughly the same number of chunks and data, avoiding hotspots.mongodb+1​
  
  

Sharding strategies (hashed vs range)

• Range sharding uses contiguous ranges of the shard key (for example, dates or IDs), which is good for range queries but needs care to avoid all new inserts landing on one shard.mongodb+1​
  
  

• Hashed sharding hashes the shard key value before assigning to a chunk, which spreads inserts more evenly across shards, especially for monotonically increasing keys.200oksolutions+1​
  
  

How data distributes in a 4‑shard cluster

• In a 4‑shard cluster, each shard owns some set of chunks; together, all shards’ chunks represent the full data set of each sharded collection.scylladb+1​
  
  

• When new documents are inserted, mongos uses the shard key, looks up the owning chunk in the config metadata, and sends the operation to the shard that currently owns that chunk.mongodb+1​
  
  

• As data grows or becomes unbalanced, the balancer splits large chunks and migrates some chunks from “heavier” shards to “lighter” shards so that all 4 shards stay close in chunk count and data size.percona+1

​

Steps to apply sharding on a database/collection

For your 4‑shard, 3‑config, multi‑mongos environment, the high‑level steps are:

1. Connect to mongos and enable sharding on the database:
   
   

   • sh.enableSharding("yourDb") enables that database for sharding.percona+1

2. Create an index on the chosen shard key fields on the target collection 

(for example, { userId: 1 } or { userId: "hashed" }).geeksforgeeks+1

3. Shard the collection:
   
   

   • sh.shardCollection("yourDb.yourColl", { userId: 1 }) for range sharding, or
     
     

   • sh.shardCollection("yourDb.yourColl", { userId: "hashed" }) for hashed sharding.mongodb+1​
     
     

4. After this, MongoDB creates initial chunks for that collection and starts placing them on the 4 shards; the balancer will keep moving chunks to maintain even distribution as data grows.bmc+1​
   
   

https://www.geeksforgeeks.org/mongodb/manage-sharded-cluster-balancer-in-mongodb/

https://devopedia.org/mongodb-sharding