Community,

I need to verify what I am seeing through my nars.  I started a case but cannot get support to call me back (7 days).  Support - if your listening it seems that the support times are getting worse.

I am seeing total LUN IOPs throughput averaging 700.  This is 87% write and 13% read.  Raid 1_0 setup.  Disk queue is 2.  Response time is well under 5ms.  Cache write hits per second is generally equal to the write IOPs so I assume these great numbers can be attributed to caching.

However the confusing part is when I compare the LUN IOPs to the disk IOPs.  If I run the numbers of each disk I would assume those numbers would be relatively close to the total LUN IOPs of all the LUNs in that RG.  This is definitely not the case.  I am seeing much lower numbers.  I have a R1_0 (3+3).  If I take each disk across each time slice and average all the numbers I am seeing total (all disk) IOPS of ~465.  Almost 30% lower than the total LUN IOPs!  Also you have to consider that other LUNs are pushing these numbers too.

My question is this.  Since these are baked in production numbers can I use the disk IOPs numbers to scale the Clariion?  Or should I go with the common knowledge of using the LUN IOPs?  It just seems like massive overkill to use the LUN IOPs since it will never use it.