VMware

CPU ready graph on the host. How big of an issue is this? Reading about it I would say quite substantial.

CPU ready graph on the host. How big of an issue is this? Reading about it I would say quite substantial.



View Reddit by ivanb141993View Source

 

To see the full content, share this page by clicking one of the buttons below

Related Articles

5 Comments

  1. I’m new with VMware and still learning about it on my current job, but I’ve noticed that our CPU Ready times on the hosts are quite high. I’m just trying to explain to my colleagues that this is something we should really look into. Since we are on the vSphere 6.0 version, after patching it, we had to disable HT because of the L1TF “esx.problem.hyperthreading.unmitigated” alarm was coming up. Any best practices/advice on what to do?

  2. expand with more hosts for cpus, right size VMs with lower cpus, and/or re-enable HT for the interim until the new cpu scheduler is improved unfortunately…I can imagine the environment is noticeably sluggish hitting those numbers 🙁

  3. This is, fundamentally, too much on too little. With virtualization, especially VMware, you are better off upsizing as needed and starting small. Get vROps in there and let it collect stats for a while, after nerfing the heck out of those huge VMs. This kind of Ready time is crippling.

    First, turn HT back on, as others have commented. For capacity numbers, only count physical cores. HT is just the gravy on top. By doing it that way, you always have some room for growth and you have time to add hosts, or add clusters, while your VMs keep humming along.

    With vROps, you can keep an eye on the actual usage within the environment (averaged but still accurate). That will tell you what can be upsized and downsized as well.

Leave a Reply