We help IT Professionals succeed at work.

Esxi VSphere 6 - how to gang servers together using Distributed Resources Management

229 Views
Last Modified: 2017-11-21
We're running vSphere 6 on nine servers, each with their own copy of vSphere installed.  Several of the servers' CPU are overloaded, while the others are idle.  I've read that a single copy of vSphere can gang together the servers into one, big processing unit and can automatically allocated CPU load from one server to another.  I believe the process was called Distributed Resource Management (DRM) in v.5.5.  How can this configuration be done in v.6?

Your help is greatly appreciated!
Comment
Watch Question

Andrew Hancock (VMware vExpert PRO / EE Fellow)VMware and Virtualization Consultant
CERTIFIED EXPERT
Fellow
Expert of the Year 2017

Commented:
DPM is the ability to power on and off hosts in a cluster.

vSphere DRS is the function of DRM...

Remember that DRS is a function of a Cluster. So all you need to do is create a Cluster, edit the Cluster and enabled DRS.

and what it does is MOVE a VM automatically  using vMotion to another server, to ensure that the VM has it's resource entitlement of CPU and MEMORY..

it is still possible if you have very active VMs are not enough host resources, even with DRM enabled, you will see High CPU and High Memory alerts on hosts.
Paul SolovyovskySenior IT Advisor
CERTIFIED EXPERT
Top Expert 2008

Commented:
In most environments you will need to have shared storage in addition to what has been mentioned in the previous post.  That will allow the compute resources to move from host to host as the VM is located on storage shared by several ESXi hosts.

Author

Commented:
Thank you, both.  I found this discription of DRS, with a handy diagram.  Looks like the functionality requires an upgrade to vSphere Enterprise Pluse.

How difficult is DRS and shared storage to configure?  I found this tutorial, which doesn't seem too difficult.  The link is six years old, however, so perhaps you could tell me if it's still viable.

Also, if anyone has had experience with DRS, how well does it work?  Are there any tradeoffs?  The tutorial I linked shows a slider bar for DRS to adjust it from less-to-more aggressive, but I don't what that means.

Thank you for your patience.  My company is small and I really need help with the basics.

Author

Commented:
What's the best rough-and-ready way I can learn this vSphere setup.
Paul SolovyovskySenior IT Advisor
CERTIFIED EXPERT
Top Expert 2008
Commented:
This problem has been solved!
(Unlock this solution with a 7-day Free Trial)
UNLOCK SOLUTION

Author

Commented:
That's a big help.  If someone in IT at the office already has experience running vSphere, is it much of a challenge to get DRS running?
VMware and Virtualization Consultant
CERTIFIED EXPERT
Fellow
Expert of the Year 2017
Commented:
This problem has been solved!
(Unlock this solution with a 7-day Free Trial)
UNLOCK SOLUTION

Author

Commented:
Is there a limit to the number of hosts I can add?  At what point does it become unrealistic to keep adding old hosts and just purchase newer, faster ones?

Also, can I adjust the level at which any of the hosts operate?  That is, can I leave the CPUs on some hosts permanently more available and leave the CPU demands higher on the others?
Andrew Hancock (VMware vExpert PRO / EE Fellow)VMware and Virtualization Consultant
CERTIFIED EXPERT
Fellow
Expert of the Year 2017

Commented:
Max Hosts in a Cluster is 64.

Adding New Hosts with different CPUs and enabling EVC is a compromise, because it masks all the new CPU instruction sets!

see here

https://blogs.vmware.com/vsphere/2014/06/enhanced-vmotion-compatibility-evc-affect-performance.html

https://www.derekseaman.com/2012/09/how-much-does-evc-mode-matter-and-which.html

So the answer it depends on your environment!

Also, can I adjust the level at which any of the hosts operate?  That is, can I leave the CPUs on some hosts permanently more available and leave the CPU demands higher on the others?

Not really, and remember an ideal Cluster is to have balanced Cluster, e.g. all hosts the same!

You can "stick VMs to hosts" but then that really negatates DRS!

Author

Commented:
Thanks again, Andrew.  The first link you referenced re: EVC seems to suggest that only some, relatively minor CPU instruction sets are disabled by EVC.  Would a major function like multi-threading be affected?  If all the hosts in a Cluster are the same, are all the instructions sets upheld?

Would you explain a bit more how to "stick VMs to hosts"?  I have a really critical server/platform/application that I'd like to ensure always receives maximum CPU power on a new host, but would still like to utilize the host's CPU when the application is not demanding it.

Author

Commented:
Thank you so much!