This is a schematic of my AWS app setup
I have an ECS auto scale cluster for my front end based on Django which retrieves data from Postgres and ElasticSearch. I have bunch of Lambdas either event driven or schedule driven with some Step function orchestration for long running or linear processes.
Everything in AWS logs to CloudWatch who's logs I can ship to ES using function beat.
However I;'m struggling with any form of correlation in APM.
I'd like to look at the underlying component performance to work out bottlenecks in the app.
e.g. is my UI sluggish because my cluster is short on resources or in the the middle of autoscaling
or is my data slow because my lambda data ingest is hammering the ES cluster etc etc
How best should I configure the AWS environment to feed APM.
Do I need speared file beat/metric beats on each ECS instance?
Do I need multiple function beats one for ECS, one for EventBridge Lambdas one for step function lambdas with custom index names?
APM is OK ish for performance tuning ES queries but I'm struggling to configure what is a fairly simple use case in APM