Cassandra internals and performance was presented. The key points covered include:
1) Cassandra has a layered architecture with APIs, a Dynamo layer, and a database layer. The Dynamo layer implements the Dynamo paper and handles replication and failure handling.
2) The database layer includes the memtable, SSTables, commit log and more. It handles writes, flushes, compactions and reads from storage.
3) A number of performance tests were shown measuring the impact of configuration parameters like memtable flush queue size, commit log sync period, and secondary indexes on write and read latency. Bloom filters, compactions and concurrency were also discussed.