Back to Article

  • ktwebb - Saturday, May 16, 2009 - link

    "How many of us are running more than 50 to 100 VMs, which need on average only 1GB per VM?"

    600+ VM's. Less than 10% run with more than a gig of memory. 90% of the VM's runnning windows 2003
  • JohanAnandtech - Saturday, May 16, 2009 - link

    On one physical server? that would probably be a new record :-). Reply
  • joekraska - Thursday, May 14, 2009 - link

    I run a virtual data center with over 800 virtual machines in it and a sustained 25+% growth rate. I found it curious that one poster wrote that "no one pays attention to this." We do pay attention to it, it's the first (and only) benchmark that I look at in deciding things on our side, however not for the reason that might first appear. To wit, we do not use it to decide between vendors: we already have the 'best' vendor, and in truth given two different vendor systems based on the same underlying intel equipment, the results are insufficiently different to compel a vendor-switch, generally speaking. Besides that, we're homogenous. No, that's not why we're interested at all.

    However, we /are/ interested in using VMMark data to help us decide which of our current vendors many platforms we will switch to, when we switch, as well as slightly influencing WHEN our recap occurs. You have to keep in mind that VMWare licenses are quite costly. And the maintenance is costly! There is a cold calculus to ongoing consolidation involving server costs, memory footprint (e.g., can the new server hold many more dimms than the last), and the price one pays for vmware. VMMark data and the ratio of memory per CPU in a proposed new server type both play a key deciding role here.


  • duploxxx - Sunday, May 10, 2009 - link

    I'll give some ideas on reality and this is based several months perf testing and many installations.

    lets say we use 4 types of VM's: dbs + file server + compression server + iis server.

    If I run these Vm's in a 2cpu 4GB config at 60-70% of load.
    All these combined on a 2s quad machine we run 3VM if we would use 4vm we see a perf drop of approx 10% in total netw and i/o.

    that is reality, quite a difference then these benchmarking scores. but the vmmark still gives a good idea about cpu performance. For example compare barcelona against harpert and the harpert will require about 10-15% more clock for equal perf, against shanghai this raised even to 20-25% as you would clearly see in vmmark. The 3,3harper was not able to perform like a 2,7 shangh.
    Now with nehalem the stakes are changed, I didn't check a full build yet but from what i've seen it will be 10-15% more shanghai speed required against nehal. HT ain't doing nothing here when you have some heavy Vm's, when you run 15-20/1 server farm you might have some added value here.

    The solution for AMD will be Istanbul, although not much faster clock/clock then shangh it will add more real cores, with current cache system and ht link the istanbul will scale good enough. For high load VM this will be a much better choice then a HT core. As long as VM off course stays in the same price/socket system :)

    That will be for sure my choice of building rig.
  • has407 - Saturday, May 09, 2009 - link

    "The easiest way to see that VMmark is showing its age is in the consolidation ratio of the VMmark runs. Dual CPU machines are consolidating 8 to 17 tiles. That means a dual CPU system is running 102 virtual machines, of which 85 are actively stressed! How many dual CPU machines have you seen that even operate half that many virtual machines?"

    True. OTOH, the more VM's, the greater the stress on the hypervisor, which is essentially a measure of how well the hypervisor multi-tasks (or more accurately "multi-VM's"). While obviously not indicative of real-world workloads, as a synthetic measure it would seem a reasonable first-order indication of the hypervisor's efficiency dealing with disparate VM's.

    That said, I'd agree that there is no substitute for measuring and analyzing real-world workloads. Anyone who depends on a VMark score for determining their virtualization strategy or how many VM's they can shove into a box is in for a rude surprise. Not to mention that VMware has quite a few papers on specific apps that provide good clues and a good basis for analysis independent of VMark.

    Then again, I don't know of anyone with more than a room temperature IQ that depends on VMark or uses it for anything but bragging rights. The typical and appropriate steps are to (1) look at the current workload; (2) do an estimate of what that will look like when virtualized; (2) try it with a subset of the worload; (3) correct estimates and calculations as needed; (4) move forward.

    In short, VMark may be flawed, but the flaw has less to do with VMark and more with the way people use it. VMark scores provide one potentially useful data point--but only one data point.
  • JohanAnandtech - Saturday, May 09, 2009 - link

    "the greater the stress on the hypervisor, which is essentially a measure of how well the hypervisor multi-tasks (or more accurately "multi-VM's").

    Excellent comment... VMmark is indeed overemphasizing relatively simple world switches.

    "While obviously not indicative of real-world workloads, as a synthetic measure it would seem a reasonable first-order indication of the hypervisor's efficiency dealing with disparate VM's."

    Indeed. But IMHO, VMmark spends too much time in the scheduler, which makes the percentage it's spends in the exception handling (Hardware virtualization works with exceptions: Memory management, Interrupts etc.) uncharacteristically low. Further analysis needed of course.
  • has407 - Saturday, May 09, 2009 - link

    Just as with any OS, the effectiveness of the scheduler is a function of how fairly and efficiently it manages the available resources. For a CPU-intensive workload it's generally pretty simple. It's when you start throwing in IO (or QoS) that really tests the mettle of a scheduler.

    The reason why those VMark reports tend to show such a ludicrously high number of tiles is because that is typically the number of tiles needed to saturate the CPU. Which is another way of saying that either: (1) the workload is too heavily IO-bound; or (2) the IO subsystem is too slow relative to the workload.

    Which is the reason the VMark tests typically ramp up the number of tiles until CPU saturation, otherwise you're testing--and constrained by--the IO subsystem's performance, (which is typically the limiting factor), and not the hypervisor's performance. Obviously the hypervisor's IO efficiency is a factor, but if you've still got 20% CPU idle, you can push it further.

    That's not to say VMark is "correct"; at the risk of oversimplification...

    1. Would a smaller number of tiles/VM's produce less time in the hypervisor? Yes, as there are fewer VM's with timers, etc. to be emulated. That could be measured by running an increasing number of idle VM's and measuring the hypervisor cost. E.g., if basic emulation takes 1% of the CPU per idle VM, then 50 VMs should consume 50% of the CPU.

    2. Would a smaller number of VM's incur less scheduling overhead? Maybe. It largely depends on what the VM's are doing:

    a) For a CPU-bound workload, the difference is probably nominal for any but an extreme number of VM's. That could be easily determined by running an increasing number of CPU-intensive VM's and calculating the additional pure scheduling overhead/VM. E.g., if the scheduling and context switch overhead is 1ms/VM, and the scheduling quantum is 100ms, then pure scheduling and context-switch overhead will consume 50% with 50 VM's

    b) For an IO-bound workload, there will be more events likely to invoke the scheduler, and thus increasing scheduler time. Is that increased scheduler time a function of the number of VM's or the number of IO events? Hard to say, but I'd argue the latter. For the VM doing the IO, that's also more likely to lead to that VM stalled waiting on IO (unless of course that VM also has CPU-bound processes), so the more VM's the more likely the hypervisor can find a non-stalled VM and keep the CPU utlized doing useful work.

    3. Would a smaller number of VM's that produced an aggregate IO load (IO events and thus scheduling events) equivalent to a bunch of VMark tiles show an appreciable difference? Maybe... too many factors to make blanket statements, such as the type of IO, the speed of the hardware IO subsystem, and the hypervisor's efficiency in performing specific types of IO. Which of course is why people really should measure their workloads. (VMWare also has a number of micro-benchmarks available that provide clues independent of VMWark.)

    In short, I think that if VMark has a problem, it's less a matter of how many VM's, and more that the aggregate IO load from all those (small) VM's is higher than what a real-world set of (larger) VM's will produce given the the number of VM's that can fit into the same amount of memory. That would mean lower CPU utilization (more likely all VM's are waiting on IO), or it means your IO subsystem is too slow (and EMC is ready and waiting to help you fix that :).
  • tviceman - Friday, May 08, 2009 - link

    Synthetic benchmarks are a worthless waste of time and reading. When I purchase a new CPU, GPU, RAM, HD, etc. for improved performance it is based only on real world applications. Reply
  • has407 - Friday, May 08, 2009 - link

    "One of the reasons that these configurations obtain higher scores is the fact they run with two DIMMs per channel, so the DDR3 DIMMs run at 1066MHz. That is good for a boost of 5-6%, which has been confirmed by both Intel's and AMD's VMmark experts. The second and most important reason..."

    You forgot about EPT support that showed up with Nehalem, which can provide nominal to very significant performance improvements.
  • JohanAnandtech - Saturday, May 09, 2009 - link

    I mention it elsewhere in the article, but I agree that I should have talked about EPT in that sentence too :-).

    EPT is probably good for about 20%. I would try it out myself, but VMmark requires at least 72 GB on a Nehalem Xeon, and that is out of the reach of our lab right now.
  • duploxxx - Sunday, May 10, 2009 - link

    time for some OEM beta hardware :) Reply
  • has407 - Saturday, May 09, 2009 - link

    With a single socket (assuming quad core), I'd think you should be able to do it with less memory, maybe 32-48GB? (unless your IO subsystem is slow) Even if that's beyond reach, a relative measure with a smaller number of tiles might be interesting (ok, it won't be strictly by-the-VMark-book). Reply
  • tshen83 - Friday, May 08, 2009 - link

    So,I take it that you want to discredit VMmark as being a relevant benchmark for virtualization?

    You know VMmark isn't the only benchmark that says Nehalem is twice as efficient in performance/watt than Shanghais right?

    In the last paragraph, you said "give us a few more days". To do what? To selectively choose a few benchmarks that show that Shanghai is a better CPU for virtualization workloads? Good luck with that.

    Sometimes I want to find a ruler and measure just how deep you stuck your head up AMD's rear end. Sometimes I also wonder why a Belgian is so freaking adamant about AMD. Anand got too cheap I guess to outsource an important job to Belgium I guess.
  • 7Enigma - Monday, May 11, 2009 - link

    What I find so funny about this post is that since the Nehelem launch the author has chronically been labeled pro-Intel in the comments section of the majority of his articles. Just goes to show you; you can please all of the people some of the time, some of the people all of the time, and in the case of Intel/AMD, none of the people all of the time. :) Reply
  • whatthehey - Friday, May 08, 2009 - link

    Or perhaps he just has some information on a new virtualization benchmark suite, which may or may not show Shanghai in a better light.

    I think it's pretty easy to conclude that the two year old design of VMmark is aging and not as relevant as when it first came out (if it was even truly relevant then). So let's wait a few more days, eh?

    Sometimes I want to pull out a ruler to measure just how far up Intel's ass tshen83 has shoved his head so that he can't even consider any viewpoint that doesn't state that Intel is unequivocally the best. Seriously, look at any AMD or Intel article, and he's there espousing the virtues of Intel and trashing everything AMD does. It's not all black and white, dude... except when you get paid by Intel to do what you do, of course.
  • Viditor - Monday, May 11, 2009 - link

    My own guess is that tshen83 has become a 100% Intel ass, the 2 things have merged in this reality...:)

    I would like to see how things compare on the larger boxes though. There are an awful lot of 4 and 8 way VM machines going out there right now...
  • noxipoo - Friday, May 08, 2009 - link

    Plenty of companies have old servers that doesn't need much. NT4 to 2000 servers can easily range to those numbers at most corporations.
  • JohanAnandtech - Saturday, May 09, 2009 - link

    Possible, but still an exception. Windows 2000 and NT4 servers have become a minority, probably less than 5% of the installed base.

    And you are probably not too concerned about CPU performance when consolidating those servers.
  • duploxxx - Friday, May 08, 2009 - link

    Finally some good article to breach this new vmmark scores.

    Although it is clear that the new Nehalem based system is better then current shanghai this is mostly due to the 3 mem controllers (which in the end provides more mem/cpu) and faster memory. The HT feature is the main VMmark whoop score cause here, it is already stated by many Vmware performance representatives that people should take care about the HT core as a real core in production, if you do this the performance will get bad just as previous HT, although the ESX sw is no much more aware of this feature (esx 3.5u4 and esx4), but is seems like the vmmark is not able to see the difference since there is not enough load on the system.

    All other features are now equal while shanghai switching time was way better then harpertown the nehalem is more or less equal, also the ept/npt or rapid V or whatever you want to call it is now implemented.

    so a final vmmark performance score you stated around 16-17 sounds very reasonable.

    the performance enhancements in esx4 are not really for HT rather the core coherency features like vmware wants to call this, iommu which will be first introduced by amd istanbul and most important the paravirtualized scsi driver and off course more cpu/vm and a lot of memory, scheduling improvement.....etc....

    Perhaps you should contact there are aware of this VMmark real world difference and are working on a new version.
  • lopri - Friday, May 08, 2009 - link

    Yet again from Johan. Johan never disappoints! I have just had a quick read, but I will take a thorough read later. Thank you much and I'd like the follow-up articles very much, too. Reply
  • JohanAnandtech - Sunday, May 10, 2009 - link

    Thanks for the compliment and especially the confidence!

    - Johan
  • tynopik - Friday, May 08, 2009 - link

    one problem with VMmark is that it tries to reduce a complex combination of variables to ONE NUMBER

    that's great if everyone has the SAME WORKLOAD, but they don't

    how about more focused benchmarks that stress ONE particular area of vm performance?

    then people can looks at the numbers that most impact them

    also it would be interesting to see where the gains are coming from

    is nehalem better at virtualization simply because it's a faster cpu? or are the vm-specific enhancements making a difference? inquiring minds want to know
  • mlambert - Friday, May 08, 2009 - link

    No one pays attention to this. It's all about whats available from HP in terms of blades & enclosures when it's time for your 3yr hardware refresh.

    Right now it's BL2's the 495's. This is how corp IT works.

Log in

Don't have an account? Sign up now