Compaction Backlog Choking Read Performance
warningWhen pending compaction tasks accumulate, read latency degrades because queries must scan many SSTables. This manifests as rising P99 read latency while write latency remains stable, often accompanied by increasing SSTables-per-read counts.
Monitor cassandra_compaction_tasks_pending staying elevated or climbing over time, correlated with increasing cassandra_client_request_read_time_99p and cassandra_sstable metrics showing high SSTable counts per table.
Increase compaction_throughput_mb_per_sec (default 16 MB/s is too low for SSDs; use 128+ MB/s). If CPU-bound, verify concurrent_compactors doesn't exceed physical cores minus 1-2 reserved for reads/writes. For LCS workloads falling behind, consider switching to STCS or adding nodes to distribute compaction load.