I read through Cassandra and BigTable sections, and have a feeling that their write operations are highly similar (write to commit log and MemTable, then periodically merge into SSTable, etc).
Could anyone elaborate why BigTable is a CP system while Cassandra is an AP system? IMO, it is because
- BigTable doesn’t do hinted handoff like Cassandra when the node/tablet is down, so it would be unavailable during this time, hence no A from the CAP.
- it serves read requests through Tablet server (which is like a single leader in its responsible key ranges), while Cassandra serves read requests from any R available nodes which may or may not have the latest data.
Is this understanding correct?