Most VMware admins know by now that auto tiering storage systems should have SIOC configured differently than if they have storage systems that don’t do auto tiering. Especially if it’s auto tiering at block level. It’s also recommended to have IO metric collection disabled because it will get different results based on where the blocks on the datastore physically reside. What you may not know is that VMware has enabled IO metric collection for every datastore by default with ESXi 6.0. This can cause some weird latency spikes on your storage system that can be very hard to isolate if you don’t know what you’re looking for or what might be causing it. So here is how you fix the problem.
If like a lot of people you still primarily use the C# client for vSphere then you will have to switch to a web browser to fix this problem. This will require you log into the vSphere Web Client. Once you are logged in select Storage -> <Datastore> -> Manage. You should see something similar to the following image:
Click on the Edit button and then click the check box next to Disable Storage I/O statistics collection as in the image below:
Click Ok and repeat the process for your other datastores.
If you would like to do the above more automatically by using a PowerCLI Script, please check out this great blog post by Burke Azbill here.
Hopefully the above helps you reduce any unexpected latency spikes on your datastores and storage systems you use with ESXi 6.0. As the number of storage systems with auto tiering and multiple storage types increases over time this becomes even more relevant. Given the trends in the industry it is a wonder VMware made this a default setting in the first place. As storage systems become more intelligent there is less and less need for features such as Storage IO Control, which means admins should be able to focus on other more important things, rather than worrying about noisy neighbor problems. FYI, If you are a Nutanix customer this also applies to you. I found this little gem while running an NCC check on a cluster that I was getting periodic unexplained latency spikes. Needless to say the problem has now been resolved after making this change.
This post first appeared on the Long White Virtual Clouds blog at longwhiteclouds.com. By Michael Webster +. Copyright © 2012 – 2016 – IT Solutions 2000 Ltd and Michael Webster +. All rights reserved. Not to be reproduced for commercial purposes without written permission.