Databases · Question 27 of 50
What is the difference between replication and sharding?
Scaling reads against scaling writes.
The answer
Replication keeps copies of the same data on several machines: a primary takes the writes and replicas copy them, so reads can be spread out and a replica can take over if the primary fails. Replicas usually lag a little, so a user may not see their own change if you read from a replica right after it; send those reads to the primary.
Sharding splits the data across machines by a key (user id, region), so each holds a part. It scales writes and storage, but queries and transactions across shards get hard, and a poor key makes hot shards. Usually you replicate first, and shard when one primary can no longer take the writes or hold the data.
The picture
replication: [primary] ── copies ──▶ [replica] [replica] same data, more reads
sharding: users 1–1M ─▶ [shard A]
users 1M–2M ─▶ [shard B] different data, more writesHow do you pick a shard key?
Show the answer
Something most queries filter by, with many values spread evenly, usually the user id. Not the creation date: every new write would land on one shard.