Application metrics also change: usually 2ms, but every 10th request — 300-400ms. Resources are fine. The app only makes requests to Redis and RDS. What could be the reasons?

The reasons for periodic response time increases (every 10th request — 300-400 ms) with normal resources and using only Redis and RDS could be:

• Cache misses in Redis: some requests may not hit the cache, leading to database access, which is slower.
• Peak load on RDS: even with normal resources, there could be locks or long queries due to contention or complex queries.
• Connection pool issues: if the connection pool to the database or Redis is limited, some requests wait for a connection to be freed.
• Garbage collection (GC) in the application: periodic GC pauses can slow down request processing.
• Background jobs or periodic tasks: such as cache updates, backups, or other operations affecting delays.

For diagnostics, it is recommended to:

• Analyze Redis and RDS logs and metrics.
• Check the usage and configuration of connection pools.
• Review GC metrics and application profiling.
• Track whether slow requests coincide with any background tasks.