We have around 3K connections to our bucket. Is there any way to determine from the sever side top N sources (IP addresses) for KV requests in the specified time frame
I’m not the expert here, but client ip addresses might be recorded in the server’s memcached.log
This doesn’t directly answer your question but have you looked at using OpenTelemetry? You can have the SDKs send spans from all operations (or a random representative sample) to an OpenTelemetry platform like Jaeger or Honeycomb. These spans will contain the node that created them, along with useful information such as how long each request took.