Problem: etcd performance gets slower when
* The request object is very large(close to 1MB). Data replication and object transmit across leader and followers is slower
* There is more and more data. Especially after running clusters over 10 years and the data size is close to 8GB.
Kubernetes issue link:
https://github.com/kubernetes/kubernetes/issues/122799 proposed solution:
* Enable compression(snappy or zstd or extended ones) on object to shrink the size(object size shrink to 40%~8% of original size by models, CRD has better compression rate)
(there could be duplication because i couldn't get the previous item in the conversation list)