Cross-Stack Performance Tuning
Latency and throughput degradation caused by interacting services and configurations.
Identify the top contributors to p99 latency including affected endpoints, downstream dependency timing, and any recent change or configuration context.
2. Which queries or operations are consuming the most resources?Surface the most resource-expensive queries or operations ranked by CPU time, I/O wait, or memory consumption; include affected data store and upstream callers.
3. What is the current saturation bottleneck type for each saturating service?Classify current saturation by bottleneck type (CPU-bound, memory-bound, I/O-bound, connection-bound, queue-bound) for each saturating service; include evidence from wait events, queue depths, worker/thread counts, and connection pool metrics.
4. How have key metrics changed across the most recent release boundary?Compare key latency, throughput, and error-rate metrics across the most recent release boundary; include config or dependency diffs that may explain regressions.
5. Which configurations are likely amplifying saturation or latency under current workload?Identify configurations whose current values are likely amplifying saturation or latency under observed workload patterns. Examples: app concurrency vs. database pool size, cache TTL vs. database load, JVM heap vs. container memory limit, HPA targets vs. request concurrency.