SCVMM/PRO Complexity: High

About a year ago, I wrote about the complex requirements for enabling PRO Tips in System Center Virtual Machine Manager 2008 (SCVMM).  That was the previous release — what has changed now that SCVMM 2008 R2 is available?

Not much.

First Things First

Before you can use PRO to Live Migrate virtual machines between hosts (which is sort of the whole point), you need to first configure Hyper-V, failover clustering, and Cluster Shared Volumes — something eWeek found to be significantly more complicated than VMware ESX.

After Hyper-V and Live Migration are up and running, the next step is to install SCVMM 2008 R2 so you can integrate it with your System Center Operations Manager 2007 (SCOM) environment.

What’s that?  You don’t have SCOM deployed?  Go ahead and set it up.  Come back in a few days when you’re ready to go to the next step.

Integration Time

After your environment has both SCVMM and SCOM up and running, it’s integration time! If you have ever thought that the next-next-finish Microsoft wizards were boring, then this is the task for you.  You’ll get a chance to make configuration changes to various systems and possibly even Active Directory — haven’t you always wanted to learn about the SetSPN utility anyway?

But don’t let me spoil the excitement by giving away the ending — pick up your own copy of the 34-page integration guide and experience the thrill for yourself.

SCVMM-SCOM integration guide

Second Opinion

Some of you are probably thinking that this is an exaggeration, it can’t be all that bad.  Let’s take a look from another perspective, presented in the Infrastructure Planning and Design (IPD) guide for SCVMM 2008 R2:

IPD-SCOM integration complexity

As you can see from the above excerpt, integrating SCVMM with SCOM clearly increases the time, effort, and complexity of your virtualization deployment.  And don’t forget that SCOM is required for monitoring your Hyper-V infrastructure, too.

Unlike the Microsoft virtualization conglomeration, VMware DRS is an integral part of vCenter Server — purpose-built for virtualization.  Talk about additional layers

Tags: , , , ,

19 comments

  1. Stu Fox’s avatar

    So taking your page number argument to it’s logical extension I should probably look at the DRS guide. The section of the resource management guide that covers DRS is 27 pages long, or putting it another way it’s still 80% as complicated as PRO. If I cut out the table of contents from the PRO guide DRS is probably 90% as complicated as PRO. Which is a silly argument, as is your page number argument.

    In terms of complexity, the meat of that document is summarised as follows:
    – Make sure the SPN is set (this is an Operations Manager best practice, and Operations Manager will alert you if it isn’t set). Setspn is familiar to most Windows admins who have done anything with SQL or IIS over the years.
    – Import the SQL & IIS Management Packs (most admins would probably have this already)
    – On your VMM server, install the OpsMgr console
    – On your OpsMgr server, insert the VMM disk & run configure Operations Manager
    – Set the PowerShell execution policy
    – Add the VMM user to the OpsMgr console
    – Enable PRO

    What you also fail to mention is that once you’ve done this, you have a far smarter tool for monitoring your physical and virtual environment. DRS treats the VM as a black box, so gives you no context about why the machine is being moved (for instance, why is your box consuming so much CPU resource? Is there a problem? OpsMgr & VMM work together to be smarter than that. And then partners come along and extend the functionality of PRO so that we can do even more with it.

    You know where I work…

    1. Eric Gray’s avatar

      Hello Stu,

      Interesting, but it looks like a bit of hand-waving.

      Would you please provide a specific example of a scenario where application-related events/errors are combined with Hyper-V VM resource utilization statistics to make a super-smart load balancing decision?

      I look forward to learning more about this capability.

      Eric

    2. Stu Fox’s avatar

      Eric

      Why should you make a “super-smart load balancing decision”? Why wouldn’t you just make a “I’ll fix this problem” decision instead – for instance resolve the problem that is causing high CPU utilisation rather than just move the machine around (which just moves the problem). That’s why having a solution that is also looking at what is going on in the VM is so powerful. So PRO might not kick in at all, but the additional functionality that OpsMgr is bringing to the table would (so for instance, the SQL admins might get an alert to say “fix your stuff”, or the BizTalk admins might get an alert to say “BizTalk is sucking memory/CPU”)

      Cheers

      Stu

    3. Aaron’s avatar

      The Microsoft bashing gets a little tiring to read after a while; how about just some positive stuff about VMware products, much more interesting.

    4. pennino’s avatar

      I don’t think this is Microsoft bashing. This is the difference between reality and some PowerPoint slides where you put vmware on one side and hyper-v on the other.

      You know, what the demos, the webcasts and the white papers do not tell you.

      You have no idea how difficult is to have non technical people (often, decision makers) understand these things without clear examples like this.

      Eric, please keep bringing them on.

    5. Fernando’s avatar

      Stu,

      For DRS to work, it is just a matter of setting up your cluster. Than, a checkbox will enable it.
      Setup your hosts, shared storage, network, and you are ready to go.
      For Hyper-V , in addition to setting up, you need many additional and complex steps.

      And your comparing is not apples to apples: DRS is not intrusive inside the VM itself (yet), you are talking about a different problem here, which indeed OpsMgr has something. But it is not integrated with ProTips.

      Also, DRS is hots based load balancing, and will not move a problem to another hosts. It will alleviate the load on a host which might be exhausted, by moving the VMs to another host.

      If your SQL is bad configured and eating all CPU cycles, this is another history.

    6. Eric Gray’s avatar

      Stu,

      You did not answer my question, which is very revealing.

      We still cannot tell why SCVMM + SCOM is so much smarter than DRS.

      Eric

    7. Stu Fox’s avatar

      Eric – It’s revealing that you’re trying to pretend that all PRO does is load balancing. PRO stands for Performance & Resource Optimisation, and one of the things it can do is load balancing. However, it does more than that as well. For instance, if a guest VM is consuming too much CPU but not stressing the host, PRO will advise you to increase the number of virtual CPU’s allocated to that guest. Or if your IIS server is stressed, PRO can provision another server for you.
      Also, as noted above PRO is partner extensible so anyone can write a PRO management pack to take other actions based on application specific information in OpsMgr (see Dell, HP, IBM, Emulex, F5, Citrix and others).
      And then as pointed out above, you also get all the functionality that is in OpsMgr out of the box as well so you can provide application level intelligence about your VM’s.

      Fernando – I wasn’t the one doing the comparison here, Eric was. And to your point that OpsMgr having something that is not integrated with ProTips – that is entirely what PRO provides – integration between VMM & OpsMgr. And alleviating load on a host is a good thing, but it might not solve the underlying problem – if the problem is that SQL is eating CPU wouldn’t you rather solve that than move the problem? Moving might be a good temporary solution but not a good long term solution.

    8. Fernando’s avatar

      Stu,

      What I mean is that, PRO does not move a VM based on what is happening inside of it (an SQL server problem for ex).

      I agree with your statements, but we are talking about 2 different problems here. One thing is making sure the hosts are providing the best possible performance to VMs. Another thing is a bad application inside a VM, a poor sized VM, a VM related performance problem, etc etc This is something that tons of solutions exists for years with monitoring software (Tivoli, Unicenter, etc etc).

      This kind of application specific monitoring is not specific to VMs. Again, we are talking about 2 different problems here. DRS is NOT designed for that.

      What Eric is comparing is apples to apples: PRO with DRS, and on this point, there’s nothing that shows PRO is better in any way. Quite the opposite, Eric shows on other posts here that PRO, is not so “PRO” 😀

    9. Fernando’s avatar

      Being more specific, the comparison is between DRS and the similar functionality which PRO delivers. I know PRO delivers other types of functionality but is another kind of monitoring, and not a DRS-like stuff.

    10. Stu Fox’s avatar

      Fernando

      OK, I get it now. If you ignore everything else that PRO does and just focus on the stuff that it does that is exactly the same as DRS, it’s no better than DRS?

      Because PRO is leveraging information from OpsMgr, any information that OpsMgr is collecting can be used as a target for PRO tips. So there is a PRO tip that triggers on guest CPU utilisation which doesn’t move the machine (what’s the point of doing that?) but instead suggests increasing the vCPU allocated to that machine. I think you’re confusing does not with cannot – just because we don’t prompt to move the machine based on guest CPU utilisation (again, why would we do that, guest CPU util won’t get resolved by moving it – a host based tip would move the machine if that also happened to be stressing the host) doesn’t mean we can’t do that. If you had a situation where wou wanted a guest counter to trigger a migration, PRO could absolutely do that.

      Cheers

      Stu

    11. Fernando’s avatar

      Hey Stu,

      I am not ignoring the features, I acknowledge it does much more , but it because it is a product with a much wider scope, not only virtualization management.

      That’s why I say the comparison is not apples to apples.
      I am not saying OpsMgr is bad, or anything like that.

      So, if a customer wants to replace ESX with Hyper-V, they need a huge complex monitoring system to have workload balancing, with lots of features it may not need.

      I know many customers have already their established monitoring tools (Tivoli, Unicenter, Openview, etc etc). Will they need to fully implement a complex system to have a DRS-like functionality ? Or throw away their existing systems ? This is the complexity Eric is talking about. Why not create a virtualization focused product un-tied from a huge monitoring system which requires many different components, database, etc etc ?

      I think the subject here is not DRS versus PRO, but how complex is to achieve the same functionality VMware gives you.

    12. Phil’s avatar

      Just to be clear, Im very much for Hyper-V. We run Hyper-V R2 in production at 15:1 consolidation with SCVMM R2 and SCOM and Hyper-v itself has demonstrated itself to be rock solid. At Edu pricing, it costs us pennies instead of the £0000s VMWare wanted for a comparable setup.

      That said, SCVMM/SCOM integration was an absolute pig to set up. And having reviewed the features DRS offers, Id rather have that (DRS) than the SCVMM/SCOM setup. It seems to be trying too hard to be a solution for the wrong problem. To me it doesnt seem too complicated to have a simple application that checks the CPU usage of VMs on the host (as exposed through WMI perfomance counters already) and balance machines based on simple rules.

      Unless Im missing something obvious, with SCOM I can’t keep VMs seperate for continuity, they can end up on the same host, thus if that host dies it will take out both VMs for a short time while they boot up on another host.

      Hopefully, Citrix will pick up on this need and add it to Essentials for Hyper-V now that Live Migration is available.

    13. Stu Fox’s avatar

      Phil – if you want to keep VM’s separate for continuity that is a function of the cluster itself. Check this KB for information: http://support.microsoft.com/default.aspx?scid=kb;EN-US;296799

      Cheers

      Stu

    14. Eric Gray’s avatar

      Is there also a way to keep two Hyper-V VMs together on one host?

    15. Alberto’s avatar

      Stu,

      you have got to be kidding me….you call that KB a solution? I can’t believe you guys are really telling people to use the command line to change the values of a registry key in order to set affinity rules and then compare PRO Tips with DRS. Come on, you can’t be serious about it. Plus, why would anyone want to set affinity rules at the cluster level anyway, when it is much more logical to set it a the VM level since that’s what you want to control to begin with? I don’t want to think what happens the moment you have more than 2 VMs to manage that way….good luck.

    16. David’s avatar

      I agree, we’ve been using Vmware (ESX and Vsphere) for years. Enabling DRS is an easy task. Enabling Pro pack takes too many ressources. I think Microsoft has just forgot the medium and small Businesses in the loop.

      I will not run SQL server to manage 3 to 5 hosts. Unfortunately, SQL 2008 is required to run SCOM.

      Installing SCVMM is not that hard. Although SCVMM is not as good as Vcenter (deploying a Vm from a template takes hour) it’s not a bad software. But none of our customers will switch from Vmware to Hyper-V without some kind of DRS.

      When Microsoft will add PRO pack or whatever it is to SCVMM, there we’ll go with Hyper-V if it still less expensive than Vmware. As Vmware Microsoft could have a licensing system to enable PRO pack on one software insteading of having another software to manage.

      Too bad, for now we’ll stick with Vmware, a little more expensive, but so wasy to manage.

Comments are now closed.