Explanation about How CPU Limit and CPU Reservation can Slow your VM (if you don’t do a proper sizing and analysis)

In this post, I would like to share about CPU limit and CPU reservation configuration in vSphere ESXi virtualisation technology.

Actually those features are great (since the configuration also available in vCloud Director (*it will call the configuration in vCenter)). Those features are great if you really know and already consider on how to use it properly. For example, if you would like to use CPU reservation please make sure that you are not running those VMs in a fully contention/overcommitment environment. For CPU limit, if you have application that is always consume 100% of CPU even though you always give more CPU to the VM – then you can use Limit configuration to limit the usage of the CPU by that application (but, for me the Best Way is ask your Developer to Fix the Application!).

Okay, let’s talk more about CPU Limit.

Duncan Epping and Frank Denneman (both are the most respectable VMware blogger), once said that: “Look at a vCPU limit as a restriction within a specific time frame. When a time frame consists of 2000 units and a limit has been applied of 300 units it will take a full pass, so 300 “active” + 1700 units of waiting before it is scheduled again.”

So, applying a limit on a vCPU will slow your VM down no matter what. Even if there are no other VMs running on that 4 socket quad core host.

Next, let’s talk more about CPU Reservation.

Josh Odgers (another virtualisation blogger) also explained that CPU reservation “reserves” CPU resources measured in Mhz, but this has nothing to do with the CPU scheduler. So setting a reservation will help improve performance for the VM you set it on, but will not “solve” CPU ready issues caused by “oversized” VMs, or by too high an overcommitment ratio of CPU resources.

The configuration of Limit and Reservation are done outside the Guest OS, so your Operating System (Windows/Linux/etc) or your Application (Java/.NET/C/etc) do not know that. Your application will ask the resource based on the allocated CPU to that VM.
You should minimize the use of Limit and Reservation as it makes the operation more complex.

Conclusion:

Better use the feature of default VMkernel which already got a great scheduler functionality that will take fairness into account. Actually, you can use CPU share configuration if you want to prioritise the VM other than others.

But, the most important thing is: “Please Bro…, Right Size Your VM!”

 

Kind Regards,
Doddi Priyambodo

 

Be Sociable, Share Tweet about this on TwitterShare on LinkedInShare on FacebookEmail this to someone