Results 1 to 12 of 12
  1. #1
    Join Date
    Jan 2006
    Location
    Sydney, Australia
    Posts
    251

    How can I work out whether a Virtuozzo UBC host is overselling?

    Hi I am posting this as a VPS customer. I have a Virtuozzo VPS using UBC, with 384MB guaranteed and around 6GB (!!) burstable, while the physical server has 8GB RAM and 12GB swap. I can see physical server's memory usage as meminfo is not virtualised, and the server constantly has less than 1GB free + buffers + cached, while the swap is currently at more than 5GB. I can only see how much memory my VPS is using -- 160MB oomguarpages and 200MB privvmpages -- so it is far below the limit of the package, but the VPS can feel slow sometimes, which I suspect might be due to swapping.

    The question I would like to ask is -- is it possible to pin-point why the VPS is slow? My apps are obviously not CPU bounded, and as iostat is not really working in Virtuozzo I cannot tell whether it is IO-bounded inside my VPS either. Is it possible to find out whether it is due to excessive paging on the physical server, so I can go to my provider and say, "hey you are overselling and you should not pack that many VE into a physical server".

    Something I suspected, but please correct me if I am wrong as I am no expert in Virtuozzo, is that high burstable amount is actually bad. I guess from host's point of view they set it high so their customers can host Java applications, but if the customers host only the LAMP stack, they can now use far more than their "guaranteed" memory as allocated/committed ratio is low.

    One question I have is -- when will OOM kick in to kill processes? When physical memory is exhausted before paging out? Or when all virtual memory is exhausted, i.e. even the swap is full. I guess it might be later? Wouldn't it be too late in my situation, that when OOM handler kicks in to kill processes in VE using more than their guaranteed memory, the entire physical server (+ ALL the VE inside) is already crawling because there is simply too much paging activity going around? In another word, other VE on the same physical server can severely affect my VE's performance and their processes are not afraid to be killed (say using "only" 5GB privvmpages and 3.5GB commited), as long as there is still swap space left?

    It could be in this situation where while my provider DOES NOT oversell based on their guaranteed memory calculation, with high burstable amount people are in fact using more than they should and drag everyone's performance down due to swapping.

    I would like to get opinions from providers who are familiar with Virtuozzo UBC/OpenVZ, and hopefully be able to push forward a recommendation to my provider so they can adjust their settings.

    Thanks.

  2. #2
    Join Date
    Oct 2006
    Location
    New York, NY
    Posts
    1,034
    Quote Originally Posted by ylsy View Post
    Hi I am posting this as a VPS customer. I have a Virtuozzo VPS using UBC, with 384MB guaranteed and around 6GB (!!) burstable, while the physical server has 8GB RAM and 12GB swap. I can see physical server's memory usage as meminfo is not virtualised, and the server constantly has less than 1GB free + buffers + cached, while the swap is currently at more than 5GB.

    I don't know about other providers, but to me, when a VPS node is using ANY swap at all that's your first warning sign that the server is being pushed to its extremes. But again, this is how I run our servers I'm not sure how others choose to do so.

    If what you say is true about the physical node having that much swap space in use, than that's your problem. When a server uses 5 GB of swap space for normal operations, you can bet that it's doing alot of I/O because of the constant paging and this will result in a general "slowness" feel to the system. Even if the underlying disk system is SA-SCSI having that much swap being used is going to slow everything down.

  3. #3
    Join Date
    Feb 2006
    Location
    New York, NY
    Posts
    704
    Also, bear in mind. If the server is using say 7.9 out of its 8GB of RAM, it's definitely just cache. Linux does use a highly advanced cache system. It will fill up its ram with cache, and it will be recorded in overall system memory. This memory is not being actively used, and can be replaced with active data at any time.
    Last edited by r00ter; 02-28-2008 at 09:42 PM.

  4. #4
    Join Date
    Jan 2006
    Location
    Sydney, Australia
    Posts
    251
    Quote Originally Posted by r00ter View Post
    Also, bear in mind. If the server is using say 7.9 out of its 8GB of RAM, it's definitely just cache.
    Yeah. Note I have said 1GB free + buffers + cached. Maybe I should have used braces - that is 1GB total of (free + buffers + cached) so it is actually not much, and 7GB RAM + 5GB swap are actually being "used".

    I checked the swappiness from inside my VE and it is 60 which is default (Question: is that value accurate from inside a VE?) so I guess it is slightly biased towards swapping out inactive pages to get more room for buffers and cache. Even that there is less than 1GB of memory for disk cache from a 8GB box...

  5. #5
    Join Date
    Jan 2004
    Location
    Pennsylvania
    Posts
    939
    My take: It's vastly oversold. Even though swap *usage* is nowhere near as notable as swap *activity* when measuring performance, using 5GB of swap on a 8GB system has to have a lot of activity. In my experience on systems with 8GB of RAM usually swap activity increases to unsafe levels when around 2GB swap used.
    Matt Ayres - togglebox.com
    Linux and Windows Cloud Virtual Datacenters powered by Onapp / Xen
    Instant Setup, Instant Scalability, Full Lifecycle Hosting Solutions

    www.togglebox.com

  6. #6
    Join Date
    Nov 2005
    Location
    Michigan, USA
    Posts
    3,872
    If 5GB of SWAP is being used, I'd say they are overselling...

    Our servers rarely get over 100mb.


  7. #7
    Join Date
    Aug 2004
    Location
    Shanghai
    Posts
    1,449
    Quote Originally Posted by TheWiseOne View Post
    In my experience on systems with 8GB of RAM usually swap activity increases to unsafe levels when around 2GB swap used.
    2GB swap used? Gosh... that must be just unusable!

    When I read posts like this, I know we did the right choice to use Xen.

    Thomas
    GPLHost:>_ open source hosting worldwide (I'm founder, CEO & official Debian Developer)
    Servers & our leading control panel and our Xen VPS hosting, which are already included in Debian and Ubuntu
    Available in: Kuala Lumpur, Singapore, Sydney, Seattle, Atlanta, Paris, London, Barcelona, Zurich, Israel

  8. #8
    Join Date
    Feb 2006
    Location
    New York, NY
    Posts
    704
    Thomas,

    Just because a host uses swap does not mean it's a reflection of the virtualization platform. Just remember that when you have one too many high resource users Also, make sure the swapped information is being used, clear your RAM cache on the VPS first and see what effect it has on the swap.

  9. #9
    Join Date
    Aug 2004
    Location
    Shanghai
    Posts
    1,449
    My idea was that it's better to have each individual VPS running it's own swap space rather than having a huge global one. Having 2GB of swap usage in a single server is just impossible in our system.
    Sorry if it was not clear enough.

    Thomas
    GPLHost:>_ open source hosting worldwide (I'm founder, CEO & official Debian Developer)
    Servers & our leading control panel and our Xen VPS hosting, which are already included in Debian and Ubuntu
    Available in: Kuala Lumpur, Singapore, Sydney, Seattle, Atlanta, Paris, London, Barcelona, Zurich, Israel

  10. #10
    Join Date
    Jan 2006
    Location
    Sydney, Australia
    Posts
    251
    Quote Originally Posted by gplhost View Post
    My idea was that it's better to have each individual VPS running it's own swap space rather than having a huge global one. Having 2GB of swap usage in a single server is just impossible in our system.
    That is one reason why I prefer Xen than Virtuozzo/OpenVZ, especially when swap partitions are on separate physical disks so as long as my VPS doesn't touch swap, it won't really be affected by other VPS that have been constantly swapping. I am not saying hosts using Virtuozzo/OpenVZ are bad -- I have been with some wonderful non-overselling providers using OpenVZ and at work we do quite a bit of server consolidation using OpenVZ. Each VE is just not as isolated as VPS under Xen.

    Hate it when my usage is clearly way under the guaranteed amount, but some of my key processes are paged out (and physpages << oomguarpages)

  11. #11
    Join Date
    Jun 2006
    Location
    Ahh' UnderPants
    Posts
    1,012
    and still some providers come and write VPS market has less overselling.
    This example shows filling VMs on a node like piglings.



  12. #12
    Join Date
    Jan 2004
    Location
    Pennsylvania
    Posts
    939
    Quote Originally Posted by gplhost View Post
    2GB swap used? Gosh... that must be just unusable!

    When I read posts like this, I know we did the right choice to use Xen.

    Thomas
    I don't see how having 2GB swapped used and a swap activity of < 0.5 pages/sec is bad. Apparently you missed the point of my post where swap *activity* is the resource you really want to monitor.

    Oh well, to each their own. You are free to use whatever platform you like, but I guess you need to try to bash others in other to feel confident in your decision.
    Matt Ayres - togglebox.com
    Linux and Windows Cloud Virtual Datacenters powered by Onapp / Xen
    Instant Setup, Instant Scalability, Full Lifecycle Hosting Solutions

    www.togglebox.com

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •