HPE Storage Users Group

So in system reporter, we're monitoring for NL IOPS >75. As of a few days ago when moving VMs to this new 3PAR 7200, I'm constantly getting emails that the IOPS for all of the drives are over 75. Some are going as high as 150 IOPS.

Any ideas on what I can do...tweak, tune, etc. to help this? Is this something where adding more drives would be the ideal solution?

Currently, the system is a 2-node 7200 with two 3.5in drive shelves, 48 2TB hard drives, running RAID 6.

Add a bunch of 10k drives, have your vvols land on the FC CPG and then use AO to move cold chunklets down to your NL disks. For sizing purposes NL should be assumed to provide zero IOPS according to the design engineer I worked with recently, he only wants very cold data on NL. During production hours our NL drives average <20 IOPS each.

This was purchased by another person as a DEV box only. However, we're apparently hitting that DEV storage pretty hard to keep these IOPS up as high as they are. System reporter is showing me every hour.

2TB NL drives are ideal for large block sequential IO... perfect examples would be production or dev backup to disk, database dumps etc, or streaming video recording from security cameras, security log file consolidation/storage, most file servers.. etc... they are horrible at small block random IO, which is pretty much what every database needs, whether its PRD or DEV.

My dev strategy is to use the SAS/FC drives, mag safe, Raid 5, set size to 8+1.
Then for prod, we also use SAS/FC drives, cage safe, Raid 5, set size to max of what the cage count will allow for cage safe, currently 5+1 for us.

_________________
Richard Siemers
The views and opinions expressed are my own and do not necessarily reflect those of my employer.