Replication vs Sharding (for read scalability vs write scalability)

Question

Can someone explain why they say db replication is more ideal for read scalability while sharding is more ideal for write scalability?

From my current understanding:

replication allows read traffic to be distributed among several replicas rather than a single node, increasing read traffic. Does this not also apply to write traffic?
sharding allows a very large db node (say for example 10 TB) to be split into several smaller nodes, which allows traffic/queries to be a lot faster (since it does not need to search through 10 TB of data). Does this not apply to both read and write traffic?

itisravi · Accepted Answer · 2024-08-04 13:44:44Z

0

Sharding can make a case for both depending on the type of workload (I/O pattern). If the writes are heavy and also not happening to just a few keys of the same shard, then distributing them over a bunch of machines reduces the load per machine allowing you to pump more writes. If the read pattern is also well distributed so that it doesn't create hotspots on just a few keys, then yes you are right.

Replication can never make a case for write scalability because every update has to be sent to all the replicas. In that sense, is actually the opposite of scalable as it causes write amplification.

answered Aug 4, 2024 at 13:44

itisravi

3,5883 gold badges25 silver badges33 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Replication vs Sharding (for read scalability vs write scalability)

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related