Cassandra

Network vs Service Latency Disambiguation

info
latencyUpdated Jan 6, 2026

Client-measured latency exceeds service-side cassandra_client_request_read_time/write_time by large margins, indicating network overhead is the bottleneck rather than database processing. This misdiagnosis leads to incorrect remediation efforts.

How to detect:

Calculate delta between application-measured end-to-end latency and cassandra_client_request metrics (read_time_99p, write_time_99p). Large gaps (>50ms) suggest network latency. Use VPC Flow Logs or network monitoring to confirm round-trip times between app and Cassandra.

Recommended action:

If network latency dominates, optimize network path: use VPC endpoints (AWS), enable Multi-Region Replication for geo-distributed apps, ensure security groups don't force routing through NAT. Check DNS resolution times. If service latency dominates, optimize queries, data model, or add nodes.