How many virtual hosts should you put in a cluster?

How many virtual hosts should you put in a cluster?

How many virtual hosts should you put in a cluster?

Date: Jul 26, 2011

Advancements in virtualization technology now let you put more virtual hosts in a server cluster than ever before. But that doesn’t mean you should.

VMware vSphere 4.1 allows up to 32 virtual hosts in a cluster, and Microsoft Hyper-V R2 Service Pack 1 allows up to 16. By adding so many virtual hosts to your server cluster, however, you force your virtualization platform to make more and more capacity calculations for load-balancing purposes. And that can hurt performance throughout your virtual server cluster.

You should also avoid putting too few virtual hosts in a cluster. Doing so can eliminate the efficiencies that a virtual server cluster brings, and it can throw a wrench into your disaster recovery plans, as well.

In this video, virtualization expert Greg Shields explains how to find the magic number of hosts for virtual server cluster sizing.

RELATED: Server cluster sizing: How many VMs is too many?


Read the full transcript from this video below:  How many virtual hosts should you put in a cluster?

Hi, this is Greg Shields with Concentrative Technology. I want to spend a minute helping you figure out that magic number of hosts that you should have in your virtual cluster. If you've been working with Hyper-V version 2008 R2 , you probably now the maximum number of hosts that you can have in such a cluster is 16. For vSphere version 4.1, that maximum number is 32. But does it make sense to spec out that cluster with a maximum number of hosts? Probably not.

Obviously, a cluster is there to support the fail over needs of its virtual machines. But a cluster is also there to support those lower balancing requirements as well, we want to make sure that our virtual machines are load balanced across the host's amount of cluster. Having more hosts in a cluster, obviously provides more locations where those virtual machines can go in the case of host failure. But, at the same time, having more hosts in a cluster means that there are more calculations that are required in order to support the load balancing needs. Having too many, simply just means that there's a lot of extra work that's required to provide the best load balance for your virtual machines in your cluster.

It's been stated that for a vSphere cluster, the sweet spot is somewhere between 16 and 24 hosts per cluster. Your mileage will, obviously, very there. But this number, this 16 to 24 host sweet spot, provides a large number of locations where virtual machines can go in the case of a failover, but doesn't unnecessarily add to the amount of calculations that DRS does require. You can assume that with Hyper V, the same sort of ratio is probably the case.

If you find yourself getting too close to the maximums, you shouldn't fear splitting apart your cluster into multiple clusters. By doing so, you reduce the amount of calculations that are required for each individual cluster. While still providing for the failover and load balancing needs your virtual machines require.
 

More on Virtualized clusters and high-performance computing

  • virtual machine

    Definition - A virtual machine (VM) shares physical hardware resources with other users but isolates the operating system or application to avoid changing the end-user experience.
  • Tidy up your infrastructure with virtualization host groups

    Tip - Use these strategies for grouping virtualization host servers to make sure your data center stays organized and easy to manage.
  • What does N+1 redundancy mean?

    Answer - In data centers, N+1 redundancy is used to express the idea that a system should be designed to withstand the loss of a component.
  • stretched cluster

    Definition - A stretched cluster is a deployment model in which two or more virtualization host servers are part of the same logical cluster but located in separate geographical locations.
  • Pods fit for virtualization, but bring HA clustering challenges

    Tip - Converged infrastructure pods are a good option for virtualization, but administrators must decide whether simplicity is worth a clustering headache.

There are Comments. Add yours.

 
TIP: Want to include a code block in your comment? Use <pre> or <code> tags around the desired text. Ex: <code>insert code</code>

REGISTER or login:

Forgot Password?
By submitting you agree to receive email from TechTarget and its partners. If you reside outside of the United States, you consent to having your personal data transferred to and processed in the United States. Privacy
Sort by: OldestNewest

Forgot Password?

No problem! Submit your e-mail address below. We'll send you an email containing your password.

Your password has been sent to: