We considered Cassandra early on. After much discussion, Eric and I decided against including it. Cassandra implements a BigTable-esque API and also takes a page from the Amazon Dynamo playbook to implement ring-based sharding.
We felt that HBase and Riak more faithfully represented BigTable and Dynamo respectively. Writing about these two databases gave us a better chance to discuss the columnar and key/value genres independently.
We felt that HBase and Riak more faithfully represented BigTable and Dynamo respectively. Writing about these two databases gave us a better chance to discuss the columnar and key/value genres independently.
It was definitely a tough choice though.