Well it's a cassandra company so it's bound to exaggerate the throughput test that shows Cassandra winning but relegating its huge weakness in latency performan

Well it's a cassandra company so it's bound to exaggerate the throughput test that shows Cassandra winning but relegating its huge weakness in latency performance (Whoa!! 10ms average read latency from in-memory store...) to an "area for improvement" list at the bottom is a bit disingenuous. It's not a small issue. It's an order of magnitude difference from Voldemort, redis and even mysql at scale.