Memtables are the single largest contributor to Cassandra’s garbage collection overhead. Kamil Leśniak's talk describes the final stage of a long progression: a fully off-heap memtable implementation building on the trie memtable

Cassandra 6.0 write performance improvements from Dmitry Konstantinov the low down on how to "do less work" and "allocate less" and "contend less" and "helping the JIT"

Fedor Ryabinin getting into the depths of the Accord protocol, the first leaderless protocol deployed in a widely used production systems.

A jammed pack Cassandra Track room has started! Scott Andreas kicks it off with some great slides on "Cassandra: A Normal Database", describing how these things are getting implemented in lots of details.