I was asked why would I not use Dedupe and Compression on an ALL-Flash 4 node cluster.
I could think of a host of reasons why I would use it but can't really think of many solid reasons.
To me, one would be that you are planning on initially getting your cluster with small SSDs which you plan on swapping out later. Once your larger better disks are in then you would want to squeeze every penny out of them, thus enabling D+C.
Another reason I could think of is saving write IO amplification. But that's not really a good point since that data would be cold and VMs wouldn't need access to it. Only very large sets of data would start to impact your IOps.
What other case am I missing here?
Sincerely,
Frank