MS Products Support on VMware ESX :(

[habibalby]

Hi,

Today we had a meeting with microsoft engineers to discuss about a new infrastructure that gonna host an ESX Servers on HP BL460 in cluster mode to host Microsoft infra structure along with Exchange 2007, MOM, SMS, ect. They said cannot support Microsoft products on vmware infrastructure even with preimer support from Microsoft

Is any 1 of you guys experiance this issue with microsoft? or is there any way that Microsoft support thier infrastructure on vmware infrastructure?

And my organization cannot accept a solution without a support? how they can be convenience?

really appreciate your inputs towards this issue

[AWo]

If it really comes hard to hard grab a P2V, V2P tool, convert the machine to a physical system and look if the error/behavior persists. If so, MS has to deal with it. If not, it's maybe a problem related to virtualization and you can deal wit it.

But until it's getting hard to hard, you don't have to tell MS that it is virtual. From my experience, the support always tried to help, but they may refuse to help you further if they can't find a solution within a specific period of time.

AWo

[Patrick_Miller · Accepted solution]

Like AWo stated, just don't tell them. There is no reason for the support people, microsoft or not, to know the machine is running in a virtual environment.

MS, and all software makers for that matter, are just going to have to accept that the data center is turning into a virtual world.

[vmroyale]

Here is a link to Microsoft's official support policy:

http://support.microsoft.com/kb/897615

[mreferre]

First you need to consider that that is not the "official MS position" on supporting MS products on VMware .... it was a MS employee telling you what it is (for him) the support policy.

Having that said .. if you look at the "so called" official MS position it states that as long as you have a premier support contract they will put "reasonable business efforts" to help you. Which to me a big nothing because I challenge everyone what a "reasonable business effort" is..... is it 1 hour debug? is it 2 weeks heads down to understand what's going on? Is it "let's try to reboot and see if it fix the problem otherwise I can't help" ? What is "reasonable business effort" TO YOU ?

Point is that they do not have a REAL AND MEASURABLE support statement for virtualization (on third party tools like VMware ESX at least).

Your best friend might be either a V2P tool that was already mentioned, or a friendly engineer you might find on the line when you have a problem.

I think that they are getting smart in the sense that "not telling them it's virtual" is no longer enough .... if they ask you an output of Device Manager they would tell immediately it's a vm based on the drivers being loaded......

Massimo.

[jayolsen]

Just curious, will VMware provide support of the guests to that level with a valid support contract?

[Patrick_Miller]

I agree that eventually they will request some information that will inform them it is running in a virtual environment, but YOU don't have to offer that information up front.

To me and my organization, "reasonable business effort" means doing everything you would normally do to troubleshoot a physical box.

In a perfect world, the software vendor would have to prove it's a problem with the virtualization and their software before you moved it over to a physical box, and not the other way around. The big fear that I have, is that companies will force you to pay an additional "support fee" in order to provide support for products running in a virtual environment.

[mreferre]

> Just curious, will VMware provide support of the guests to that level with a valid support contract?

Not at all.

Massimo.

[Cloneranger]

Yes Microsoft are holding out on supporting apps on virtual servers, I have a feeling they will start supporting them as soon as they have a viable virtualization platform to sell

That said, I run all my Microsoft non-supported apps on VMs simply because missing out on Microsoft 'support' isnt something I am particularly worried about,

I have been in IT almost 11 years and in that time I have called Microsoft support about 3 times, and found them to be useless on every occasion.

[mreferre]

Patrick,

that is what YOU think "reasonable business effort" mean. This might not the same view others might have. That is the problem: subjectivity Vs objectivity. That's why all these discussions are political discussions (I have seen many SMB customers getting black eyes from MS people on the matter ...... the same MS people that says "yes sir!" to the big bank's CIO telling them "this is not open for discussion you will support us on this no matter what!"). That's what I call the difference between "asking permission to MS in order to do something" and "informing MS that you have done something". Obviously the political (and economical) weight counts in these discussions......

>In a perfect world, the software vendor would have to prove it's a problem with the virtualization and their software before you moved it over to a physical box, and not the other way around. The big fear that I have, is that >companies will force you to pay an additional "support fee" in order to provide support for products running in a virtual environment.

How is the saying? You hit the nail on the head?

That's exactly the point. It's about assumptions: We on this board assume that the virtualization layer works unless otherwise proven. MS think that their sw works unless otherwise proven. It's a matter of what you debug first to get it "out of the picture". Now if you ask me .... I think that a generic ISV position like "prove me that it's not the virtualization layer and then I will help you" is laughable. Not because I like VMware / Xen or virtualization in general .... but simply because, if you consider what happened so far ..... 99.99999999% of the problems reported to the ISV's by the users are either due to products misconfigurations or due to products bugs. Very rarely it is due to a bug in the hypervisor or its misconfiguration.

So if this is the reality .... does it make sense for every customer to prove "it's not the hypervisor" when we know that 99.9999999% of the time it's not the hypervisor? I will try to for sure to get the hypervisor out of the picture .... but only after you spent 3 weeks heads down to try to debug your problem and I will try a physical box only as one of the last debug resorts.

Testing an application on a physical box is not per se a bad thing. We know that ISV's sometimes ask you to do very pervesavive things to debug (I remember MS wanted you to un-cluster certain applications for debug or to un-bind network adapters etc etc etc)..... I think that moving to physical can be consider a step of the problem determination ..... the only problem is which step? The first one or the last resort? That is the main point we need to convince ISV's of.

Massimo.

P.S. all my discussion refers to functional problems....not performance problems. Performance is totally another thing and I am more with the ISV's on this....

[RParker]

For that matter, even on the MS page they offer Exchange, SQL 2008, and MOM on a virtual hard drive to be run on their environment. So if they don't support Virtual Servers, why is the website implying that you can virtualize these servers then?

I think they are trying to weasel out of something.

But I agree, no one has to know. A Virtual Server isn't any different than a normal Physical host.

People fear what they don't understand, that's what is really going on.

[mreferre]

> I have been in IT almost 11 years and in that time I have called Microsoft support about 3 times, and found them to be useless on every occasion.

This is another excellent point. I remember that back 7-8 years ago when I started discussing VMware deployments some customers made the point that "this is not officially supported by MS". The funny thing is that these are the same customers that have been complaining for years with MS support that was "slow to respond, could only suggest to reboot or reinstall, or if they find a bug they say it will be fixed in the next release of the product".

I have always wondered why customers gets worried about "not being" supported. Some of them just need to cover their bu*ts I think ... (I am sure this is not the case for the original poster of this thread.... no pun intended here).

Most people think that "getting support" means "assurance that someone will fix my problem". I don't agree with this assumption ..... many times I have seen customers solving problems on their own or referring to forums/blogs.... many times I have seen customers loosing arrays of data on their "supported" enteprise storage ..... with the "support center" saying "sorry you need to reload from tape". Etc etc etc

I think that, for the x86 space, we should really call "support" ...... "willing to help" instead ......... this would set the right expectations... and we would all leave in a better world.

Massimo.

[RParker]

> the same MS people that says "yes sir!" to the big bank's CIO telling them "this is not open for discussion you will support us on this no matter what!"). That's what I call the difference between "asking permission to MS in order to do something

Its better to ask forgiveness than permission. If it works, do it. If it's been working for a long time, and not now what is the problem? It's not the device, it's the people. People make changes, not devices.

> So if this is the reality .... does it make sense for every customer to prove "it's not the hypervisor" when we know that 99.9999999% of the time it's not the hypervisor? I will try to for sure to get the hypervisor out of the picture .... but only after you spent 3 weeks heads down to try to debug your problem and I will try a physical box only as one of the last debug resorts.

No it's not our job to prove their product sucks. It's our job to relay information about the performance of the product. Someone wants to always blame someone else for their mistakes. They don't want to believe that their precious product isn't perfect. Someone else thinks they have all the answers, but in reality they don't

[Patrick_Miller]

Patrick,

that is what YOU think "reasonable business effort" mean. This might not the same view others might have. That is the problem: subjectivity Vs objectivity. That's why all these discussions are political discussions (I have seen many SMB customers getting black eyes from MS people on the matter ...... the same MS people that says "yes sir!" to the big bank's CIO telling them "this is not open for discussion you will support us on this no matter what!"). That's what I call the difference between "asking permission to MS in order to do something" and "informing MS that you have done something". Obviously the political (and economical) weight counts in these discussions......

King,

I couldn't agree more...I was simply stating what I thought the practice SHOULD be, especially if you are paying for a support call or contract. But we all know the differences between what should be, and what actually is...

[Patrick_Miller]

the same MS people that says "yes sir!" to the big bank's CIO telling them "this is not open for discussion you will support us on this no matter what!"). That's what I call the difference between "asking permission to MS in order to do something

Its better to ask forgiveness than permission. If it works, do it. If it's been working for a long time, and not now what is the problem? It's not the device, it's the people. People make changes, not devices.

So if this is the reality .... does it make sense for every customer to prove "it's not the hypervisor" when we know that 99.9999999% of the time it's not the hypervisor? I will try to for sure to get the hypervisor out of the picture .... but only after you spent 3 weeks heads down to try to debug your problem and I will try a physical box only as one of the last debug resorts.

No it's not our job to prove their product sucks. It's our job to relay information about the performance of the product. Someone wants to always blame someone else for their mistakes. They don't want to believe that their precious product isn't perfect. Someone else thinks they have all the answers, but in reality they don't

Amen! (In a completely agnostic way )

This is a very complicated issue, and one I am eager to follow as we continue to move into this virtual world.

[RParker]

> Today we had a meeting with microsoft engineers to discuss about a new infrastructure that gonna host an ESX Servers on HP BL460 in cluster mode to host Microsoft infra structure along with Exchange 2007, MOM, SMS, ect. They said cannot support Microsoft products on vmware infrastructure even with preimer support from Microsoft

HOGWASH! That's is complete and utter disdainfully and irrefutably ignorant. TOTALLY.

First, doesn't Micrsoft make a Virtual Server? YEAH!

Do they offer those SAME products for virtualization on their own virtualization website? YEAH.

Ask them to prove that statement, in writting. Make them pull out the agreement and show you that it says word for word exactly that statement. That's a useless statement. Especially when Microsoft own website demonstrates that those products work PERFECTLY well on their own Virtual Server. We all know ESX is better than VS, therefore:

ESX > VS > MS products on Physical hosts.

And lets get down to nitty gritty.. exactly what does a Virtual Server do DIFFERENT than a physical one? EXACTLY. Ask that over paid, hyped up, close minded, dark ages, mainframed mind set cretan to explain the difference.. I would be glad to sit in on that one. Let him dig himself out of that hole.

[Cloneranger]

Most 'real techies' (for want of a better expression) myself included and a handful of guys I have met over the years:

a) wouldn't be caught dead calling MS or other Hardware/Software Vendor for support,

b) can solve most problems from their own knowledge anyway,

c) actually take some pride in solving their own problems, using the internet for research is part of finding your own solutions,

d) and when you really need help you have usually over the years built up your own support network of crack specialists you can talk you, guys you currently or previously worked with,

Speaking to say an exchange specialist I know for example, I would achieve more in a ten minute chat then days on the phone to MS.

[mreferre]

Amen. There is nothing to add.

Well the only thing is ... this gives me an opportunity to clarify that mine was not a bash against MS. I think these problems are horizontal in the industry rather than vertical to a specific ISV. When you buy something with a "3 years warranty included" or "sw support for 300$ a year" ... you know what you get (i.e. close to nothing).

To people that are only exposed to the x86 platform this might sound blasfemy I know ..... but people that have been exposed to other "platforms" probably know what I mean....

Massimo.

[dmorgan]

Along these same lines, what is everyone using for MS and for Linux to do V2P conversions? I understand that many vendors, MS, Oracle, etc... that we use don't officially support their products in a VM environment. That doesn't bother us, because if we can migrate that virtual machine to a physical, and reproduce the errors, then we can prove it is not the VM, and thus get support. However, for as many V2P converters I have found, I have found very few that mention V2P. Recommedations anyone?

If you found this or any other post helpful please consider the use of the Helpfull/Correct buttons to award points

[Cloneranger]

Oh I am no way knocking MS specifically either,

The sad thing is they are one of the better vendors out there,

Their phone support may not be any good, but whos is?

What is good, is the amount of work they put into their KBase, their hotfixes and service packs, I also give MS kudos for making their KBase and updates FREELY AVAILABLE, TO THE GENERAL PUBLIC,

They require no contract, no registration, no login.. nothing,

They also ACTUALLY TEST things before they release them, so many vendors are so desparate to get the jump on competiton that they release half finished products with major issues,