hello,
I have a server with 40 cpu cores and 126GB memory, I installed three elasticsearch data nodes on this server, but I found that the total cpu usage is very low. So how to inprove the cpu usage of elasticsearch, can I bound cpu cores to each data node?
thanks.
What type of workload do you have? Elasticsearch is often limited by disk I/O rather than CPU. Have you verified that you storage is not a bottleneck?
I have monitored the spu memory and disk i/o on grafana, I saw the i/o is also not high, so I think the disk i/o is not the bottleneck.
Are you having performance problems? What is the use case and workload?
I use elasticsearch as log system. my elasticsearch cluster used to installed on virtual server, each server install one node. Every day nearly 0.5 billion logs will put in to es, I found the virtual server's cpu usage was very high. But on the phyisical server with 3 data nodes, the cpu usage is very low ,40 cpu cores usage is nearly 20% , I doubt the machine resources is not fully used.
If you are not having performance problems maybe the current resource usage is all that is needed.