Kassandra - Rack Concept Concept with PropertyFileSnitch

Question

I am working on deploying multiple DCs, and one thing I don’t understand is the interpretation of the rack concept from Cassandra’s point of view.

I can enforce replication order by setting the correct key ranges. Why do I need to additionally specify racks in cassandra-topology.properties ?

Take the Cassandra documentation as an example: http://www.datastax.com/docs/1.1/cluster_architecture/replication

If I have a replication factor of 3, and my row key is stored on Node 1, then the replicas will be stored on Node 2 and 3 - this is obvious when we look at the ring structure, so ... why I need to duplicate this information in the rack configuration ?

Answer 1

The rack configuration allows cassandra to optimize replica placement so you have the best resiliency features. If you have all your replicas in rack 1 and this rack is down, you will lose data. If you tell Cassandra about your rack configuration, she will store replicas on different racks.