Balg, Andreas
2010-Aug-09 09:09 UTC
AW: [Xen-users] Impressive Performance Problems with Small-Files /Small Blocksizes
Hello Todd, I''ve allocated 2GB of memory to the guests - The host-system has 16GB of RAM and I run 7 VMs only - so all happens within the boundaries of physical RAM. I could offer you the raw data as well - I could send you an excel Sheet with all of the Graphs and data used so far. Would that be useful to you? Of course I''d be interested in your results as well. Thanks, Andreas -- ----------------------------------------------------------------------------- XiNCS GmbH MwST-Nr: 695 740 Schmidshaus 118 HR-Nr: CH-300.4.015.621-9 CH-9064 Hundwil AR Webseite: http://www.xincs.eu AGB: http://www.xincs.eu/agb.html Tel. +41 (0)31 526 50 95 ----------------------------------------------------------------------------- ----- Originalnachricht ----- Von: Todd Deshane <deshantm@gmail.com> Gesendet: Son, 8.8.2010 03:17 An: "Balg, Andreas" <a.balg@xincs.eu> Betreff: Re: [Xen-users] Impressive Performance Problems with Small-Files /Small Blocksizes Hi, How much memory is allocated to the guests? The reason I ask is that are these runs all or most in memory? Do you have raw data that I could look at? I am doing some benchmarking of my own and trying to figure out some strange KVM data. Thanks, Todd On Fri, Aug 6, 2010 at 8:06 AM, Balg, Andreas <a.balg@xincs.eu> wrote:> > hello everybody, > > during some extensive benchmarking for an evaluation we found some issues with > i/o-performance and also memory bandwidth of Xen 3.4.2 (using XCP-0.5) running > on a quite up-to-date and performant Dell-R610 Server (2 x Xeon E5620, 16G RAM, > 4 x SATA 15K HDD''s - RAID 5): > > - Especially for small block sizes (below 32k) the I/O is very poor. > > to give some figures: The same Benchmark > "time iozone -az -i0 -i1 -i2 -i8 -Rb results.xls" > > Runs around 3 Minutes on the bare Hardware, around 30 Minutes in a KVM-VM > and more than 1 hour(!) in a xen VM - See attached graphs and focus on the > front of the diagram (red and blue "foot" of the xen graph) > > What I''d like to know is, if this is be a glitch in a device driver, an error in our configuration or might be eliminated in any other way using a workaround or other version. > > Or is this a proble of the differences in Design or just nobody noticed it so far and it should be lokked at by the developers. Without these two significant problems Xen would outperform kvm in almost any possible manner .... > > > Best regards > Andreas > > > > -- > ----------------------------------------------------------------------------- > XiNCS GmbH MwST-Nr: 695 740 > Schmidshaus 118 HR-Nr: CH-300.4.015.621-9 > CH-9064 Hundwil AR > > Webseite: http://www.xincs.eu > AGB: http://www.xincs.eu/agb.html > Tel. +41 (0)31 526 50 95 > ----------------------------------------------------------------------------- > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users >-- Todd Deshane http://todddeshane.net http://runningxen.com _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Todd Deshane
2010-Aug-09 12:49 UTC
Re: [Xen-users] Impressive Performance Problems with Small-Files /Small Blocksizes
Hi Andreas, You really should run I/O tests that run outside of host memory so you can see the performance hit at the disk level. Otherwise you are just testing RAM. I would like to see your results, but I am also interested in seeing the disk performance, not just testing of the RAM. I am collecting my results here: http://github.com/deshantm/Rapid-Recovery-Desktop-Testing/tree/master/results My guest systems have 1G allocated to them. The host system has 4 GB. Desktop class system, Intel Core 2 duo 2.4 GHz. So far only KVM, but Xen is planned for future work. Thanks, Todd On Mon, Aug 9, 2010 at 5:09 AM, Balg, Andreas <a.balg@xincs.eu> wrote:> Hello Todd, > > I''ve allocated 2GB of memory to the guests - The host-system has 16GB of RAM > and I run 7 VMs only - so all happens within the boundaries of physical RAM. > > I could offer you the raw data as well - I could send you an excel Sheet with > all of the Graphs and data used so far. Would that be useful to you? Of course > I''d be interested in your results as well. > > Thanks, > Andreas > > > -- > ----------------------------------------------------------------------------- > XiNCS GmbH MwST-Nr: 695 740 > Schmidshaus 118 HR-Nr: CH-300.4.015.621-9 > CH-9064 Hundwil AR > > Webseite: http://www.xincs.eu > AGB: http://www.xincs.eu/agb.html > Tel. +41 (0)31 526 50 95 > ----------------------------------------------------------------------------- > > > ----- Originalnachricht ----- > Von: Todd Deshane <deshantm@gmail.com> > Gesendet: Son, 8.8.2010 03:17 > An: "Balg, Andreas" <a.balg@xincs.eu> > Betreff: Re: [Xen-users] Impressive Performance Problems with Small-Files /Small Blocksizes > > Hi, > > How much memory is allocated to the guests? The reason I ask is that > are these runs all or most in memory? > > Do you have raw data that I could look at? > > I am doing some benchmarking of my own and trying to figure out some > strange KVM data. > > Thanks, > Todd > > On Fri, Aug 6, 2010 at 8:06 AM, Balg, Andreas <a.balg@xincs.eu> wrote: >> >> hello everybody, >> >> during some extensive benchmarking for an evaluation we found some issues with >> i/o-performance and also memory bandwidth of Xen 3.4.2 (using XCP-0.5) running >> on a quite up-to-date and performant Dell-R610 Server (2 x Xeon E5620, 16G RAM, >> 4 x SATA 15K HDD''s - RAID 5): >> >> - Especially for small block sizes (below 32k) the I/O is very poor. >> >> to give some figures: The same Benchmark >> "time iozone -az -i0 -i1 -i2 -i8 -Rb results.xls" >> >> Runs around 3 Minutes on the bare Hardware, around 30 Minutes in a KVM-VM >> and more than 1 hour(!) in a xen VM - See attached graphs and focus on the >> front of the diagram (red and blue "foot" of the xen graph) >> >> What I''d like to know is, if this is be a glitch in a device driver, an error in our configuration or might be eliminated in any other way using a workaround or other version. >> >> Or is this a proble of the differences in Design or just nobody noticed it so far and it should be lokked at by the developers. Without these two significant problems Xen would outperform kvm in almost any possible manner .... >> >> >> Best regards >> Andreas >> >> >> >> -- >> ----------------------------------------------------------------------------- >> XiNCS GmbH MwST-Nr: 695 740 >> Schmidshaus 118 HR-Nr: CH-300.4.015.621-9 >> CH-9064 Hundwil AR >> >> Webseite: http://www.xincs.eu >> AGB: http://www.xincs.eu/agb.html >> Tel. +41 (0)31 526 50 95 >> ----------------------------------------------------------------------------- >> >> _______________________________________________ >> Xen-users mailing list >> Xen-users@lists.xensource.com >> http://lists.xensource.com/xen-users >> > > > > -- > Todd Deshane > http://todddeshane.net > http://runningxen.com >-- Todd Deshane http://todddeshane.net http://runningxen.com _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users