The production Cassandra cluster will run Jessie from the start. Currently there aren't any Cassandra packages in our Jessie repo (at least in labs), so we need to import Cassandra. We are generally using the Apache packages: http://wiki.apache.org/cassandra/DebianPackaging
We have been testing with both 2.0 and 2.1 so far. 2.1.2 [actually looks fairly stable](https://issues.apache.org/jira/browse/CASSANDRA-8029?jql=project%20%3D%20CASSANDRA%20AND%20Status%20%3D%20Open%20AND%20%22Since%20Version%22%20%3D%202.1.0%20ORDER%20BY%20priority%20DESC). It also has [some good improvements](http://www.datastax.com/dev/blog/whats-new-in-cassandra-2-1). The most relevant for us are probably the ability to tune the number of entries to cache in memory per partition for cacheable workloads (read: only keep latest revision of title in memory), incremental repair & off-heap memtables.
I think it's worth starting with 2.1 even if it means that we'll take a small risk of running into some unforeseen issue.
See also: [full changelog](https://git1-us-west.apache.org/repos/asf?p=cassandra.git;a=blob_plain;f=CHANGES.txt;hb=refs/tags/cassandra-2.1.2).