Hey there, I am trying to find the sweet spot for my application.
I got these results:
DATA nodes 8cpu-32gb ram with 16gb heap size and 8gb young heap size
GC time % is generally around 5% (this is in case the limit)
DATA nodes 8cpu-64gb ram with 16gb heap size and 8gb young heap size
GC time % is generally over 10%
In the second scenario, I doubled up the DATA's ram to try to handles more events. Instead, I am able to process lower events because DATA resources are stressed by GC time %.
All the surrounded details (input traffic, indices size, number of shards etc etc) are the same.
I read several docs on blog and also somewhere else but I cannot find a way to understand correctly this behaviour. I just know that I should find a way to decrease that GC time % value.
is there anyone that can help me? The ES version is so old, I know. I have another branch with latest one but values cannot be compared.