High Availability + Replication = Disaster

CEPH performs way too slow for our needs. We need lightning fast response times. RAID0 local seems to work the best in our testing.
 
Then maybe you get best enterprise nvme and local storage.
Yes, that’s what I did. It’s why I started this thread. I can’t get replication between nodes working with local storage. Shared storage is also not an option due to latency.