Adaptive concurreny

[shamj]

I have 3 sinks : sending metric to datadog datadog_metrics, sending logs to splunk: splunk_hec and now adding aws_s3 sink.
However when I send logs from vector sink to vector source and then aws_s3 sink , I noticed the Total log rate dropping on vector sink that's back pressuring the other transforms and sources behind it.
There is no memory , CPU max , network thruput max out. I also don't one or two errors occasionally. I also added adaptive concurrency to all three sinks. But the behavior didn't change. I am testing other keys related to adaptive concurrency , but unable to understand which one is the right one in this scenario.

Couple of questions:

• What is adaptive concurrency based on : is it cpu, memory, worker threads or something else?
• What does it do : [request.adaptive_concurrency.decrease_ratio] , [ewma_alpha] , [initial_concurrency],[max_concurrency_limit], [rtt_deviation_scale] ? (https://vector.dev/docs/reference/configuration/sinks/aws_s3/#request.adaptive_concurrency.rtt_deviation_scale)(https://vector.dev/docs/reference/configuration/sinks/aws_s3/#request.adaptive_concurrency.max_concurrency_limit)(https://vector.dev/docs/reference/configuration/sinks/aws_s3/#request.adaptive_concurrency.initial_concurrency), (https://vector.dev/docs/reference/configuration/sinks/aws_s3/#request.adaptive_concurrency.ewma_alpha)](https://vector.dev/docs/reference/configuration/sinks/aws_s3/#request.adaptive_concurrency.decrease_ratio)
  Please advise

[shamj]

Could someone please advise on this.