Field notes from operating Cassandra in production.
September 16, 2026 · RapidMindLabs
Running Cassandra on Kubernetes is now a normal choice, not a brave one — but the operator automates the easy half of the job. What cass-operator and K8ssandra genuinely solve, the storage and scheduling decisions nobody can make for you, and when staying on VMs is the right call.
Read moreSeptember 15, 2026
Cassandra 5 ships UCS, one compaction strategy with a dial that spans tiered and leveled behaviour. Here is what the parameters actually mean, how to migrate a table without a write-amplification surprise, and which tables we still leave on TWCS.
Read moreSeptember 14, 2026 · RapidMindLabs
Dual writes, backfill, shadow reads, cutover, rollback. A practical sequence for moving a production Cassandra or DSE keyspace to Astra DB without a maintenance window — and the failure modes each step is designed to catch.
Read moreSeptember 14, 2026 · RapidMindLabs
Half the Cassandra latency incidents we review are not cluster problems. They are driver defaults nobody revisited. What to set for timeouts, retry policy, speculative execution, load balancing, and paging — and how each default fails.
Read moreSeptember 14, 2026 · RapidMindLabs
Replication is not backup. A practical scheme for Cassandra snapshots, incremental backups, and commit-log archiving — plus the four restore scenarios worth rehearsing and the RPO/RTO numbers each one really gives you.
Read more