Is there a best practice on the SAN optimization of LUNs for CSV VHDX files - e.g. sequential vs. random access?
We recently set up a small two-node Hyper-V 2012 R2 Failover Cluster. As I was creating LUNs for the CSV vhdx files, our SAN (like most, I think) has some pre-set optimization options which are more or less sequential vs. random access. There's now the abstraction layer of shared VHDX files and the actual data those VHDXs are being used to store. Are there any best-practices for SAN optimization in this regard?
In other words, I could see:
A. Cluster-shared VHDXs are accessed (more-or-less) based on the type of data they're used for
B. All cluster-shared VHDXs are (more-or-less) accessed sequentially
C. All cluster-shared VHDXs are (more-or-less) accessed randomly.
I have one source that says that for a relatively simple SMB setup like we have that "C" is the recommendation. I'm curious if anyone else has run into this or seen an official best-practice...?