A alternative to cassandra for Small Projects on low end hardware?

Question

I'm preparing a infrastructure for a new project:

• data center 1, server and db
• dc 2, server and db
• dc 3, server and db
• dc x

(all quite low end servers, to reduce starting costs, so latency may be high)

The db's should form a cluster spanning the dc's. Consistency is not so important, loosing a few records when something goes down is not a problem or that dc1 db sees different data compared to dc2 db for a few seconds.

What is important is the automatic failover and correct automatic behavior when a dc comes up again.

Cassandra does all this and more, but i suspect it is a bit overkill, cause my dataset is small and my hardware isn't very good. I would only be using cassandra because of the high availability features.

Is there a alternative to cassandra which is better suited for low end/small projects but that still offers all the high availability features including replication across dc's?

Answer 1

I don't know enough about your application to know whether it would be the best choice, but I can tell you that Cassandra can be tuned to run in as little as a few hundred MB of RAM or less quite easily if your dataset is small.

Answer 2

Two DBs you can check are couchbase (http://www.couchbase.com/) and riak (http://basho.com/riak/). Both have decent cross-dc features.