For cassandra all writes must be transmitted to all replicas. CASSANDRA-1314 does not change how writes happen. Write operations will still effect cache (possibly evicting things if cache is full). Reads however will prefer a single node of it's possible replicas. This should cause better cache utilization and less duplication for those using READ.ONE with lower read repair settings.
It is also worth pointing out that the HBase cache is on entire hdfs blocks, while Cassandra can cache on keys (key cache) or a key and all it's columns (row cache). This has some deep implications based on how random your reads are. Even with Cassandra's normal cache duplication having more fined grained caches, of or rows rather then blocks, could mean that they are more efficient anyway.
