CODEWITHSUNDEEP
amazon-ec2cassandracluster-computing
user449219
user449219

Reputation: 1343

Cassandra N copies in N cluster?

I'm trying to configure Cassandra cluster on EC2.
The thing is that (for my purposes) I want to have N replicas in N machine cluster (all machine should have the same data).

I did the following:
- made a N machine cluster; all seeds; I deployed schema with replication factor N
- populate the base with WRITE ALL consensus
- now I'm trying to access data with WRITE ANY, and READ ONE.
- I load balance my clients and theoretically I should have a N time better throughput, however that is not the case.

nodetool shows in Owns column sum of 100%, instead of N*100% (each node should have all data).

any suggestions?

Upvotes: 3

Views: 237

Answers (1)

jbellis
jbellis

Reputation: 19377

If you increase replicas to N you will not see any throughput benefits, since Cassandra now has to write N copies. You will also not see any throughput benefits on reads, unless you disable read repair.

Best practice is to keep replica count constant as you increase N.

Upvotes: 4

Related Questions