ESXi 5.1 Clusters Mixing Hardware

I had a long discussing today with one of our seniro engineer regards clusters and different hardware..

Background...
We use IBM HS22/23 and Flex blades and just got our first shipment of Cisco UCS  blades today.

I suggest we created a cluster just to hold all UCS servers. this was it would cut down on vmotion issues with dissimilar hardware / CPU's..

I got over ruled and my senior staff member said there is no reason to break out he Cisco hardware into their own cluster and can mix with the existing IBM's...

Thought's.. This is not about being  right or wrong I just want learn... :-)
LVL 20
compdigit44Asked:
Who is Participating?
I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
EVC is a trade off, compromise .... so therefore something is traded....and that's performance....

EVC is designed to mix old and new hardware, and the vMotion issue goes away.....

BUT, in many Design Circles, and Design Meetings we now how, EVC is considered a last resort, and work around, or fudge, and many now believe ALL New Hardware should be Clustered together, so you can benefit from the new series of Intel or AMD CPUs.

e.g. if you introduce lets say Ivy Bridge, Sandy Bridge or Haswell architecture in your news servers, if you create a combined cluster using EVC with a Westmere EVC Baseline, you will LOSE ALL the new features of the current Intel Generation Chips!

and this will have an effect on the performance of VMs, and the new OS, because they will not be able to use the new features.

 So, right or wrong, depends on your application, but I think his thinking is OLD SCHOOL!

Over a beer, I think, that's an open debate, I would like to take part in....

some evidence and blogs here

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

http://www.vmware.com/files/pdf/techpaper/VMware-vSphere-EVC-Perf.pdf

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

(maybe you could email him these, and ask him to eat pie, if you are feelling smug!)
0
ITAddictCommented:
VMware advises against mixing dissimilar hardware (especially different CPUs, and absolutely not AMD/Intel mix), as it's bad practice when it comes to DRS, HA, FT, vMotion, etc. I suppose the "brand" like HP/Dell "might" not be an issue (if they had the same internal hardware specs/brands), although I would never do so personally. I do know you must have EVC (Enhanced vMotion Computability).

Also, I'd recommend getting off of 5.1, it has a lot of issues. I'd go to 5.5.

MORE INFO HERE:
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1005764


Let me know if you need more info. But in short, YOU are right to advise against it.
0
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
@nwembly Where is your reference to VMware advises against mixing dissimilar hardware!

When we have a VMware Hypervisor, ALL Virtual Hardware is the same!

It's - EVC is short for Enhanced vMotion Compatibility

which makes this "bad practice" compatible!

I don't believe 5.1 has any more issues than 5.5!

We are still running ESX 2.5.4 in Production! (with no issues and stable!)
0
Ultimate Tool Kit for Technology Solution Provider

Broken down into practical pointers and step-by-step instructions, the IT Service Excellence Tool Kit delivers expert advice for technology solution providers. Get your free copy now.

Mr TorturSystem EngineerCommented:
Hi,
I used EVC for some customers who are mixing ESX with different CPUs in a cluster, and there was HA, DRS, VMotion working without issues.
But as stated by A.Hancock you will loose some CPU instructions from the newer CPU, aligning to the less performant CPU if I may say so.
EVC is the easier way VMware has implemented cpu masking (since vsphere 4 if I remember correctly), and it is easier, but as cpu masking you will loose some of the new cpu benefits.

Here is one of some VMware KB talking about CPU masking :
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1993

Well, Andrew Hancock, an ESX 2.5 still in production...! Also this is not really my rule, but sometimes when it works, we got nothing to change! A proof of VMware strength and stability.
0
compdigit44Author Commented:
Thanks for the great feed back everyone...
 
In short there is no issue in creating one cluster that has a mix of DEll, IBM and Cisoc hardware using EVC. you just take a slight performance hit.

My argument was that we can vmotion between clusters and keeping host in one cluster would allow use to isolate them in some way if we need to implement cluster wide settings..

Maybe I am way off base
0
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
Thanks for the great feed back everyone...
 
In short there is no issue in creating one cluster that has a mix of DEll, IBM and Cisoc hardware using EVC. you just take a slight performance hit.

No issues...

There is not really any performance hit, unless you need to use the CPU instruction sets for applications.... So how well do you know your applications, to use these enhanced instruction set.

My argument was that we can vmotion between clusters and keeping host in one cluster would allow use to isolate them in some way if we need to implement cluster wide settings..

Well that's also possible, many VMware sites, do things differently, so have large budgets and replace the entire cluster with new hosts, some VMware Sites, add a new host each year, which has a different generation CPU, and therefore enable EVC to make the most of their investment.

Some split clusters, into "silos" for different departments, different applications, different research grants, many different ways, I don't think there is a genuine, this is the best...

However, we do split Clusters for Servers and Clusters for VDI..... and the latest EVO Rail Deployments, are for VDI ONLY!

VMware Sites do not want to impact Servers, and SLA, with Desktop Performance!
0

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
compdigit44Author Commented:
Thanks ..... great feedback everyone
0
It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today
VMware

From novice to tech pro — start learning today.

Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.