How to organize data separation between multiple databases when using sharding?

Data sharding involves dividing data across multiple databases (shards) to scale and improve performance. The main approaches to data partitioning include:

• Horizontal sharding (range-based): data is split by value ranges of a key (e.g., users with IDs 1-10000 in one shard, 10001-20000 in another).
• Hash sharding: the data key is processed through a hash function, and the result determines which shard the data belongs to. This ensures even distribution.
• Key-based sharding: the shard is directly determined by the key's value.

In Go, an abstraction layer is usually created that determines the appropriate shard based on the key and directs the request. Example:

func getShard(key string) *sql.DB {
    shardIndex := hash(key) % len(shards)
    return shards[shardIndex]
}

// Usage:
db := getShard(userID)
// Execute query on the selected shard

It is also important to consider data consistency and scalability when adding new shards.