UNSOLVED

parashuram.naya

updated

15 years ago

0

1232

July 14th, 2011 18:00

CX4-120 SAn LUNs are running with HIgh latency assigned to ESX

We have cx4-120 SAn we are using sata drives in the DAEs we did configured the LUN of 1TB each and we have total 10 LUNs assigned to 5esx servers.whenever users acces the Vms on the Datastores they found to be VMs acce3s slow and we checked the ESX logs and found high latency in Datastores.Please advise any setup we need to do in Cx box .Please let us know what best we can do to reduce letency on Luns and also make ESX environmnet more usable.we use MRU(most recent used)path for the assignment of the LUns.

  • dynamox

    11 Legend

    •

    20419 Posts

    •

    87439 Points

    342

    0

    Posted July 14th, 2011 19:00

    performance of sata is probably not cutting it , specifically for random workloads like VMware. Do you have access to Navisphere Analyzer ? If not , asked your local EMC TC to download encrypted version of performance stats (naz file) and have support analyze it for you.

  • kelleg

    6 Operator

    •

    4537 Posts

    342

    0

    Posted July 15th, 2011 09:00

    More than likely you are overloading the SATA didks. See Chapter 5 -page 95

    EMC CLARiiON Performance and Availability Release 30 Firmware Update Applied Best Practices.pdf

    http://powerlink.emc.com/km/live1/en_US/Offering_Technical/White_Paper/h5773-clariion-best-practices-performance-availability-wp.pdf

    glen

  • Nichols1

    1 Rookie

    •

    16 Posts

    342

    0

    Posted July 15th, 2011 11:00

    Here are few other things to consider ...

    1) Even though you have 10 LUNs, what is the physical disk configuration for these LUNs?  How many RAID sets do you have?  How are they assigned to the LUNs?

    2) How many vms are running on your LUN?  Remember if you have 20 vms, they can all be hitting the same physical disk.  You need less vms/disk.

    3) Although 1 and 2 may solve your problem, I agree with dynamox and kelleg.  7200 rpm SATA drives are going to be very painful for your environment.

  • christopher_ime

    6 Operator

    •

    1962 Posts

    342

    0

    Posted July 21st, 2011 21:00

    Of course as already pointed out, reviewing the array and disk layout should be your first priority; however, assuming you have done a best effort with your SATA drives and what you have to work with, if you are running ESX/ESXi 4.1, see if Storage I/O Control (SIOC) can help prioritize the more critical VM's and reduce the impact on the shared bandwidth from less critical VM's.  To borrow the header from a VMware PPT slidedesk about the "I/O Sharing Problem" that Storage I/O Control has been designed to address:

    1) Low priority VM can limit I/O bandwidth for high priority VMs
    2) Storage I/O allocation should be in line with VM priorities

    The "shares" and "limits" you set would only be applied when contention occurs and that metric is latency (30ms by default but configurable).