I created a 3 node cluster using AWS.
The we I do the test is, using 500 goroutines reading from a channel and randomly do insertion on those three nodes. And using a ticker to let 100 goroutines to feed that channel 5k messages every second. ( trying to simulate a 5k QPS)
So the result I got is the picture. No matter what I tried, the QPS on the dashboard is around 3k, and only one node seems really working.
The IOPS is only 1k, I attached two 300G EBS on the main one and using two --store when I start the server. So I don’t think that’s disk problem.
My targeting is to reach 8k QPS.