Hi everyone, I am studying the Xen performance metrics. My questions are if we can get the following metrics from Xen hypervisor interface / XEND API (XEN management API) or not ? cpu.extra.summation, cpu.ready.summation, disk.busResets.summation, disk.commandsAborted.summation, disk.queueLatency.average, mem.active.average, mem.overhead.average If no, plan to add them? or How to add them? As I study, there are the following results about XEN performance metrics on the current XEN 4.1.0. BTW, VMWARE hypervisor can get all the metrics below. We can determine if the VM / host encounter the bottleneck issues using the performance metrics. Xen (hypervisor) interface Xen enabled kernel (drivers like balloon, network / block backend, privcmd,…) XEN (management) API (XEND API) Libvirt API CPU 1. cpu.extra.summation 2. cpu.ready.summation 3. cpu.usagemhz.average v x V V disk 4. disk.busResets.summation 5.disk.commandsAborted.summation 6. disk.totalLatency.average 7. disk.queueLatency.average Throughput: An average of 8. disk.read.average and 9. disk.write.average 10. disk throughput = (disk.read.average + disk.write.average ) /2 x v V V memory 11. mem.active.average 12. mem.consumed.average v x V V 13. mem.overhead.average Memory Swapping: 14. mem.swapin.average, 15. mem.swapout.average 16. mem.swapped.average v x x V 17.mem.vmmemctl.average(balloon) x v x V network 18. net.received.average, 19. net.transmitted.average and 20. net.usage.average x v V V Thanks. YJ _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
The metric labels aren''t really descriptive enough to figure out exactly what they are. If you describe what information you want, we may be able to get it to you. Using the xend api is probably not a good plan going forward -- it''s likely that we''re going to remove xend in the next release (4.2). You may want to consider looking at XCP, which has an api, and also collects a number of statistics. If not, there is a lot of data available, it''s just a matter of getting it "plumbed" from Xen / wherever to somewhere you can use it. -George 2011/4/7 <Lu.YJ@inventec.com>> Hi everyone, > > > > I am studying the Xen performance metrics. > > My questions are if we can get the following metrics from Xen hypervisor > interface / XEND API (XEN management API) or not ? > > *cpu.extra.summation, cpu.ready.summation,* > > *disk.busResets.summation, disk.commandsAborted.summation, > disk.queueLatency.average,* > > *mem.active.average, mem.overhead.average* > > If no, plan to add them? or How to add them? > > > > As I study, there are the following results about XEN performance metrics > on the current XEN 4.1.0. > > BTW, VMWARE hypervisor can get all the metrics below. > > We can determine if the VM / host encounter the bottleneck issues using the > performance metrics. > > * * > > *Xen (hypervisor) interface* > > *Xen enabled kernel (drivers like balloon, network / block backend, > privcmd,…) * > > *XEN (management) API * > > *(XEND API)* > > *Libvirt API * > > *CPU* > > *1. cpu.extra.summation * > > *2. cpu.ready.summation * > > *3. cpu.usagemhz.average * > > v > > x > > V > > V > > *disk* > > *4. disk.busResets.summation * > > *5.disk.commandsAborted.summation * > > *6. disk.totalLatency.average * > > *7. disk.queueLatency.average * > > *Throughput: An average of * > > *8. disk.read.average and * > > *9. disk.write.average * > > *10. disk throughput* > > * = (disk.read.average + * > > *disk.write.average ) /2* > > x > > v > > V > > V > > *memory* > > *11. mem.active.average * > > > > > > > > > > *12. mem.consumed.average * > > v > > x > > V > > > > V > > > > *13. mem.overhead.average * > > > > > > > > > > *Memory Swapping: * > > *14. mem.swapin.average, * > > *15. mem.swapout.average * > > *16. mem.swapped.average * > > v > > x > > x > > V > > *17.mem.vmmemctl.average(balloon) * > > x > > v > > x > > V > > > > *network* > > > > > > > > > > *18. net.received.average, * > > *19. net.transmitted.average and * > > *20. net.usage.average * > > x > > v > > V > > > > V > > > > > > > > Thanks. > > YJ > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel > >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Hi George, The more detail descriptions about performance metrics I have is the enclosure which comes from vkernel. http://www.vkernel.com/resources/whitepapers/20-vm-performance-metrics? As I know, the XCP is a xen-based open src SW to build private/public cloud like VMWARE vcloud. And xm/XEND will be fully replaced by xl/libxl on XEN 4.2.0. So, do you mean the XCP uses xl/libxl to provide XEN management API? I can get the latest XEN related docs “XEN management API, user manual and XEN interface” “ from XEN 4.1.0 src code using command “make –C docs”. Does the XEN management API doc apply to xl/libxl? If no, where to get the latest doc about xl/libxl? Thanks YJ From: dunlapg@gmail.com [mailto:dunlapg@gmail.com] On Behalf Of George Dunlap Sent: Friday, April 08, 2011 6:20 PM To: Lu, YJ (盧盈志 TAO) Cc: xen-devel@lists.xensource.com Subject: Re: [Xen-devel] Xen Performance metrics The metric labels aren''t really descriptive enough to figure out exactly what they are. If you describe what information you want, we may be able to get it to you. Using the xend api is probably not a good plan going forward -- it''s likely that we''re going to remove xend in the next release (4.2). You may want to consider looking at XCP, which has an api, and also collects a number of statistics. If not, there is a lot of data available, it''s just a matter of getting it "plumbed" from Xen / wherever to somewhere you can use it. -George 2011/4/7 <Lu.YJ@inventec.com> Hi everyone, I am studying the Xen performance metrics. My questions are if we can get the following metrics from Xen hypervisor interface / XEND API (XEN management API) or not ? cpu.extra.summation, cpu.ready.summation, disk.busResets.summation, disk.commandsAborted.summation, disk.queueLatency.average, mem.active.average, mem.overhead.average If no, plan to add them? or How to add them? As I study, there are the following results about XEN performance metrics on the current XEN 4.1.0. BTW, VMWARE hypervisor can get all the metrics below. We can determine if the VM / host encounter the bottleneck issues using the performance metrics. Xen (hypervisor) interface Xen enabled kernel (drivers like balloon, network / block backend, privcmd,…) XEN (management) API (XEND API) Libvirt API CPU 1. cpu.extra.summation 2. cpu.ready.summation 3. cpu.usagemhz.average v x V V disk 4. disk.busResets.summation 5.disk.commandsAborted.summation 6. disk.totalLatency.average 7. disk.queueLatency.average Throughput: An average of 8. disk.read.average and 9. disk.write.average 10. disk throughput = (disk.read.average + disk.write.average ) /2 x v V V memory 11. mem.active.average 12. mem.consumed.average v x V V 13. mem.overhead.average Memory Swapping: 14. mem.swapin.average, 15. mem.swapout.average 16. mem.swapped.average v x x V 17.mem.vmmemctl.average(balloon) x v x V network 18. net.received.average, 19. net.transmitted.average and 20. net.usage.average x v V V Thanks. YJ _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel