Hello -
I'm doing some research for a potential proof of concept we're looking into. However before we purchase storage for the PoC, I want to make sure I fully understand the underlying technology. Today we provision two basic types of volumes, RAID-5's for data storage, and RAID-10's for heavy read/write volumes. These are simple 1:1 VMFS to LUN mappings on our SAN. This has worked pretty well thus far, but I recently did some reading on RAID-50/60 and thought it had some compelling advantages. The only downside is that our current storage system doesn't supported nested raid levels (besides 10).
I was able to find some reading online about VMFS "extents", but it all seemed pretty old and more focused on older versions of the VMFS file system. So my question to the community is -
Say I have a 24-disk shelf. If my SAN supported nested RAID levels, I could create a 24-disk RAID-50, which would be 4x 6-disk RAID-5's. In most cases, this will yield a much lower write-penalty than if I did 2x 12-disk RAID-5's or, a huge 24-disk RAID-6. In terms of read & write performance, how would this differ from creating a single VMFS volume that spanned 4x individually created RAID-5's? I can't seem to find any documentation about how VMWare spreads the data across the LUNs under the VMFS.
Also in general, does spanning a VMFS volume across multiple LUNs seem like a good idea? Bad idea? Horrible idea?. I can't seem to find much data on the web for this.