Interesting report on comparison of a limited number of databases. Data seems a bit simple and academic. Real world problems would show a larger difference. I personally don’t believe the rates achieved in this document. ‘Eventual consistency’ should be treated with scepticism as its simply not credible in an Enterprise installation. What I was most surprised about was MongoDBs poor performance even with transactions disabled. Also, why didn’t they test at more than 4000 requests/sec for MongoDB? Could it not handle it? I’m still poring through the report, but it doesn’t look good for MongoDB or 10gen.
I ran across an interesting and comprehensive read this morning for database junkies like me, A vendor-independent comparison of NoSQL databases: Cassandra, HBase, MongoDB, Riak, thanks to the folks at Altoros Systems Inc. and NetworkWorld. Check it out if you are interested in very thorough analysis, evaluations, and comparisons of:
- Cassandra
- HBase
- MongoDB
- Riak
- MySQL Cluster
Tons of data and charts to pour through, and some good conclusions on how to help you choose the solution that fits your specific needs.
Cheers, Steve
Adam's Deep Technology Blog 