HPE Storage Users Group

Hi all,

Following situation:

Software

HP 3PAR 7200 Release Version 3.1.3 (MU1)

two cluster nodes with 3PAR Hyper V 2012 R2.

On one virtual machine (Generation 2) OS Windows Server 2012 R2 with CSV Volume Raid 1

We have started several performance tests with CrystalDiskMark to check disk read and write.

Also we check performance with another volume (IBM Storage DS 4800) on the same virtual machine. The IBM Storage DS 4800 Storage are seven years old !

Volume on 3PAR is RAID 1 FC disks
Volume on DS is RAID 5 FC disks


The performance SEQ Read test on the DS Volume is much better as
on the 3PAR volume ! Write performance is marginally better on 3PAR.

I cannot explain why the performance on 3PAR Volume is so bad !

I try also disabling ODX on Host and virtual machine, but the result is the same .

Has anyone ideas what we can do ?

How we can show the standard cache settings for the 3PAR ?

3PAR is better on random IOs.
This array is made for random IOs.

Use a proper benchmark tool, not one intended for desktops.
http://www.iometer.org/doc/downloads.html

Also you should be testing random IOps, multiple threads and large queue depths. The system isn't designed to go as fast as possible in a straight line e.g sequential throughput. It's intended for consolidation of multiple mixed workloads all running in parallel. Make sure you are testing against a large enough data set, also have a look at what the 3PAR is actually doing from the IMC performance charts. You'll probably find it's actually sat idle waiting for some work to do, because you aren't feeding it enough requests.

There's really not enough information to go on. A 7 year old array with 200 15K RPM disks is probably going to be faster than a new disk array with 20 10K RPM disks. If you can provide more information about the configurations of the two arrays, it is more likely that you would get useful information.