This forum is now a read-only archive. All commenting, posting, registration services have been turned off. Those needing community support and/or wanting to ask questions should refer to the Tag/Forum map, and to http://spring.io/questions for a curated list of stackoverflow tags that Pivotal engineers, and the community, monitor.
I'm a heavy Cassandra user (plus created Roo) and I'm not entirely sure how you'd marry the two in a way that would deliver significant out-of-the-box functionality to standard enterprise application development.
Working with Cassandra in a non-trivial use case requires you to consider up-front your queries and then design your column families accordingly. Once you have column families designed, you're in a position to perform your Java coding with something like Hector. You can happily code Roo applications which use Hector, and indeed once Roo 1.2 comes out you'll have far more flexibility for DAO and services layer support to make this easier.
Fundamentally Cassandra differs from noSql approaches like MongoDB which are closer to a "relational" design model in so far as you design the data structure first (in this case a document) and can query it later. You setup indexes to make this more achievable. This is a more natural fit for something like Roo because we're accustomed to thinking about enterprise applications in terms of tables or entities and these are easily modelled as documents.
Having said that I have many nice things to say about Cassandra if you have serious big data requirements and are happy to trade some querying flexibility for the many benefits of a column database model. In my experience Cassandra is very pleasant to work with assuming you don't have ad-hoc querying requirements.