Aurora MySQL intermittent performance issues

Hi Everyone,

We have been noticing some intermittent performance issues with Aurora MySql. In our architecture, we have app servers and use ec2(r5.2x large) setup on autoscaling and we have roughly 4x instances running during peak hours. For transactional DB we have Aurora MySQL dedicated instance (db.r5.8xlarge) setup.

The problem I am facing here is that the 95th and 99th percentile response time of simple GET API requests in Newrelic is sometimes 10-15s which is causing timeouts on external services that are using these APIs

For reference, I have added a screenshot of such an API which is throwing timeouts intermittently.

I then went further and looked into the APM trace of Newrelic for these transactions and looks like most of the time is consumed in opening the DB connection. Screenshot attached.

The trouble is that I am not exactly sure which metric I can track in ec2/aurora to replicate or get to the bottom of this issue.

Any thoughts/guidance would be appreciated!