• 14
name

A PHP Error was encountered

Severity: Notice

Message: Undefined index: userid

Filename: views/question.php

Line Number: 191

Backtrace:

File: /home/prodcxja/public_html/questions/application/views/question.php
Line: 191
Function: _error_handler

File: /home/prodcxja/public_html/questions/application/controllers/Questions.php
Line: 433
Function: view

File: /home/prodcxja/public_html/questions/index.php
Line: 315
Function: require_once

Looking at our performance figures on our existing VMware ESXi 4.1 host at the Datastore/Real-time performance data

  • Write Latency
    • Avg 14 ms
    • Max 41 ms
  • Read Latency
    • Avg 4.5 ms
    • Max 12 ms

People don't seem to be complaining too much about it being slow with those numbers. But how much higher could they get before people found it to be a problem?

We are reviewing our head office systems due to running low on storage space, and are tossing up between buying a 2nd VM host with DAS or buying some sort of NAS for SMB file shares in the near term and maybe running VMs from it in the longer term.

Currently we have just under 40 staff at head office with 9 smaller branches spread across the country. Head office is runnning in an MS RDS session based environment with linux ERP and mail systems. In total 22 VMs on a single host with DAS made from a RAID 10 made of 6x 15k SAS disks.

      • 2
    • If head office is running on a single host you should probably get a second one regardless of storage latency ;)
    • My understanding is that average latency for a VMFS datastore should be 20ms or less, so I don't see a problem here.
    • @phoebus - We are aware of the redundancy issues of having a single host and that will play a part in our actual decisions, but that is not my question.

Definitely agree with both answers already provided here.

If head office is running on a single host you should probably get a second one regardless of storage latency - phoebus Nov 1 '13 at 12:17

My understanding is that average latency for a VMFS datastore should be 20ms or less, so I don't see a problem here. – joeqwerty Nov 1 '13 at 15:12

I would dig deeper just to get an understanding of where the latency is coming from so you have an overall big picture of where you could improve on any potential latency issues.

The values you listed are not all that bad and especially if no one is complaining. Are you currently using DAS or NAS storage? Also have you noticed any specific VM that could be causing any Latency issues?

You can dig deeper using ESXTOP from the CLI and get an idea of your DAVG, QAVG, KAVG and GAVG metrics:

enter image description here

The VMWare blog has a good explanation of these.

  • 6
Reply Report
      • 2
    • Any idea which of DAVG, QAVG, KAVG, GAVG metric if any is being reported in the vSphere Client performance tab for datastore read and write latency? It's going to take me a while to figure out the large amount of info that ESXTOP gives out (9015 columns).
    • @BeowulfNode42 I can't find references on this, but would infer from the drawing provided in the VMWare blog that the storage adapter and storage path metrics are DAVG (where a storage adapter might average over more than one path) while the "Virtual Disk" latency of the VM is GAVG. The Datastore latency probably would also include either QAVG or KAVG but as these values are really small, my observations show that the latency values reported for the datastore are mostly identical to the ones of the corresponding storage path(s).
      • 2
    • After watching the figures for a while and comparing them to user complaints on our environment, an average latency on either read or write of 15ms to 20ms is the borderline, and anything worse is definitely bad.

Trending Tags