We have cluster with 3 dedicated master nodes, 4 dedicated ingest nodes and 28 dedicated data nodes. Data sent to ingest nodes from filebeat. Max.performance indexing on this configuration ~200000 tps.
Question: if it has reason use dedicated coordinator nodes before ingest nodes and sent data from filebeat to coordinator? Or scale dedicated nodes?
Welcome to our community!
That's not really worth doing as it'll add another step to the process. I'd add more ingest nodes if you need to scale more.
Thanks a lot, Mark!