Re: [vpp-dev]: Trouble shooting low bandwidth of memif interface
Hello Rajiv, I don’t have any documentation about fine tuning performance on multicore in lxc. VPP 20.05 significantly improved mq support for tapv2. It also improved support for GSO which may be useful depending on your use case. Regards, Jerome De : Rajith PR Date : vendredi 31 juillet 2020 à 10:07 À : Jerome Tollet Cc : vpp-dev Objet : Re: [vpp-dev]: Trouble shooting low bandwidth of memif interface Thanks Jerome. I have pinned VPP thread to different cores (no isolation though). And yes after migrating to tapv2 interface, performance significantly improved (75x times approx). Do you have any documentation on how to fine tune the performance on multi core in an lxc container. Thanks, Rajith On Thu, Jul 30, 2020 at 5:06 PM Jerome Tollet (jtollet) mailto:jtol...@cisco.com>> wrote: Hello Rajith, 1. are you making sure your vpp workers are not sharing same cores and are isolated? 2. host-interfaces are slower than vpp tapv2 interfaces. Maybe you should try them. Jerome De : mailto:vpp-dev@lists.fd.io>> au nom de "Rajith PR via lists.fd.io<http://lists.fd.io>" mailto:rtbrick@lists.fd.io>> Répondre à : "raj...@rtbrick.com<mailto:raj...@rtbrick.com>" mailto:raj...@rtbrick.com>> Date : jeudi 30 juillet 2020 à 08:44 À : vpp-dev mailto:vpp-dev@lists.fd.io>> Objet : Re: [vpp-dev]: Trouble shooting low bandwidth of memif interface Looks like the image is not visible. Resending the topology diagram for reference. [cid:image001.png@01D66AB0.7BFF8520] On Thu, Jul 30, 2020 at 11:44 AM Rajith PR via lists.fd.io<http://lists.fd.io> mailto:rtbrick@lists.fd.io>> wrote: Hello Experts, I am trying to measure the performance of memif interface and getting a very low bandwidth(652Kbytes/sec). I am new to performance tuning and any help on troubleshooting the issue would be very helpful. The test topology i am using is as below: Erreur ! Nom du fichier non spécifié. Basically, I have two lxc containers each hosting an instance of VPP. The VPP instances are connected using memif. On lxc-01 i run the iperf3 client that generates TCP traffic and on lxc-02 i run the iperf3 server. Linux veth pairs are used for interconnecting the iperf tool with VPP. Test Environment: CPU Details: *-cpu description: CPU product: Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz vendor: Intel Corp. physical id: c bus info: cpu@0 version: Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz serial: None slot: U3E1 size: 3100MHz capacity: 3100MHz width: 64 bits clock: 100MHz capabilities: x86-64 fpu fpu_exception wp vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb invpcid_single pti ssbd ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid mpx rdseed adx smap clflushopt intel_pt xsaveopt xsavec xgetbv1 xsaves dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp md_clear flush_l1d cpufreq configuration: cores=2 enabledcores=2 threads=4 VPP Configuration: No workers. VPP main thread, iperf client and server are pinned to separate cores. Test Results: [11:36][ubuntu:~]$ iperf3 -s -B 200.1.1.1 -f K -A 3 --- Server listening on 5201 --- Accepted connection from 100.1.1.1, port 45188 [ 5] local 200.1.1.1 port 5201 connected to 100.1.1.1 port 45190 [ ID] Interval Transfer Bandwidth [ 5] 0.00-1.00 sec 154 KBytes 154 KBytes/sec [ 5] 1.00-2.00 sec 783 KBytes 784 KBytes/sec [ 5] 2.00-3.00 sec 782 KBytes 782 KBytes/sec [ 5] 3.00-4.00 sec 663 KBytes 663 KBytes/sec [ 5] 4.00-5.00 sec 631 KBytes 631 KBytes/sec [ 5] 5.00-6.00 sec 677 KBytes 677 KBytes/sec [ 5] 6.00-7.00 sec 693 KBytes 693 KBytes/sec [ 5] 7.00-8.00 sec 706 KBytes 706 KBytes/sec [ 5] 8.00-9.00 sec 672 KBytes 672 KBytes/sec [ 5] 9.00-10.00 sec 764 KBytes 764 KBytes/sec [ 5] 10.00-10.04 sec 21.2 KBytes 504 KBytes/sec - - - - - - - - - - - - - - - - - - - - - - - - - [ ID] Interval Transfer Bandwidth [ 5] 0.00-10.04 sec 0.00 Bytes 0.00 KBytes/sec sender [ 5] 0.00-10.04 sec 6.39 MBytes 652 KBytes/sec receiver --- Server listening on 5201 -
Re: [vpp-dev]: Trouble shooting low bandwidth of memif interface
Thanks Jerome. I have pinned VPP thread to different cores (no isolation though). And yes after migrating to tapv2 interface, *performance significantly improved (75x times approx)*. Do you have any documentation on how to fine tune the performance on multi core in an lxc container. Thanks, Rajith On Thu, Jul 30, 2020 at 5:06 PM Jerome Tollet (jtollet) wrote: > Hello Rajith, > >1. are you making sure your vpp workers are not sharing same cores and >are isolated? >2. host-interfaces are slower than vpp tapv2 interfaces. Maybe you >should try them. > > Jerome > > > > > > > > *De : * au nom de "Rajith PR via lists.fd.io" > > *Répondre à : *"raj...@rtbrick.com" > *Date : *jeudi 30 juillet 2020 à 08:44 > *À : *vpp-dev > *Objet : *Re: [vpp-dev]: Trouble shooting low bandwidth of memif interface > > > > Looks like the image is not visible. Resending the topology diagram for > reference. > > > > > > > > > > On Thu, Jul 30, 2020 at 11:44 AM Rajith PR via lists.fd.io rtbrick@lists.fd.io> wrote: > > Hello Experts, > > > > I am trying to measure the performance of memif interface and getting a > very low bandwidth(652Kbytes/sec). I am new to performance tuning and any > help on troubleshooting the issue would be very helpful. > > > > The test topology i am using is as below: > > > > [image: Image supprimée par l'expéditeur.] > > > > Basically, I have two lxc containers each hosting an instance of VPP. The > VPP instances are connected using memif. On lxc-01 i run the iperf3 client > that generates TCP traffic and on lxc-02 i run the iperf3 server. Linux > veth pairs are used for interconnecting the iperf tool with VPP. > > > > *Test Environment:* > > > > *CPU Details:* > > > > *-cpu > description: CPU > product: Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz > vendor: Intel Corp. > physical id: c > bus info: cpu@0 > version: Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz > serial: None > slot: U3E1 > size: 3100MHz > capacity: 3100MHz > width: 64 bits > clock: 100MHz > capabilities: x86-64 fpu fpu_exception wp vme de pse tsc msr pae > mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse > sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp constant_tsc art arch_perfmon > pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq > dtes64 monitor ds_cpl vmx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid sse4_1 > sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand > lahf_lm abm 3dnowprefetch cpuid_fault epb invpcid_single pti ssbd ibrs ibpb > stibp tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 avx2 > smep bmi2 erms invpcid mpx rdseed adx smap clflushopt intel_pt xsaveopt > xsavec xgetbv1 xsaves dtherm ida arat pln pts hwp hwp_notify hwp_act_window > hwp_epp md_clear flush_l1d cpufreq > configuration: cores=2 enabledcores=2 threads=4 > > > > *VPP Configuration:* > > > > No workers. VPP main thread, iperf client and server are pinned to > separate cores. > > > > *Test Results:* > > > > [11:36][ubuntu:~]$ iperf3 -s -B 200.1.1.1 -f K -A 3 > > --- > Server listening on 5201 > --- > Accepted connection from 100.1.1.1, port 45188 > [ 5] local 200.1.1.1 port 5201 connected to 100.1.1.1 port 45190 > [ ID] Interval Transfer Bandwidth > [ 5] 0.00-1.00 sec 154 KBytes 154 KBytes/sec > [ 5] 1.00-2.00 sec 783 KBytes 784 KBytes/sec > [ 5] 2.00-3.00 sec 782 KBytes 782 KBytes/sec > [ 5] 3.00-4.00 sec 663 KBytes 663 KBytes/sec > [ 5] 4.00-5.00 sec 631 KBytes 631 KBytes/sec > [ 5] 5.00-6.00 sec 677 KBytes 677 KBytes/sec > [ 5] 6.00-7.00 sec 693 KBytes 693 KBytes/sec > [ 5] 7.00-8.00 sec 706 KBytes 706 KBytes/sec > [ 5] 8.00-9.00 sec 672 KBytes 672 KBytes/sec > [ 5] 9.00-10.00 sec 764 KBytes 764 KBytes/sec > [ 5] 10.00-10.04 sec 21.2 KBytes 504 KBytes/sec > - - - - - - - - - - - - - - - - - - - - - - - - - > [ ID] Interval Transfer Bandwidth > [ 5] 0.00-10.04 sec 0.00 Bytes 0.00 KBytes/sec > sender > [ 5] 0.00-10.04 sec 6.39 MBytes 652 KBytes/sec > receiver > --- > Server listening on 5201 > --- > > > > > >
Re: [vpp-dev]: Trouble shooting low bandwidth of memif interface
Hello Rajith, 1. are you making sure your vpp workers are not sharing same cores and are isolated? 2. host-interfaces are slower than vpp tapv2 interfaces. Maybe you should try them. Jerome De : au nom de "Rajith PR via lists.fd.io" Répondre à : "raj...@rtbrick.com" Date : jeudi 30 juillet 2020 à 08:44 À : vpp-dev Objet : Re: [vpp-dev]: Trouble shooting low bandwidth of memif interface Looks like the image is not visible. Resending the topology diagram for reference. [cid:image001.png@01D66676.6DEB4E80] On Thu, Jul 30, 2020 at 11:44 AM Rajith PR via lists.fd.io<http://lists.fd.io> mailto:rtbrick@lists.fd.io>> wrote: Hello Experts, I am trying to measure the performance of memif interface and getting a very low bandwidth(652Kbytes/sec). I am new to performance tuning and any help on troubleshooting the issue would be very helpful. The test topology i am using is as below: [Image supprimée par l'expéditeur.] Basically, I have two lxc containers each hosting an instance of VPP. The VPP instances are connected using memif. On lxc-01 i run the iperf3 client that generates TCP traffic and on lxc-02 i run the iperf3 server. Linux veth pairs are used for interconnecting the iperf tool with VPP. Test Environment: CPU Details: *-cpu description: CPU product: Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz vendor: Intel Corp. physical id: c bus info: cpu@0 version: Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz serial: None slot: U3E1 size: 3100MHz capacity: 3100MHz width: 64 bits clock: 100MHz capabilities: x86-64 fpu fpu_exception wp vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb invpcid_single pti ssbd ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid mpx rdseed adx smap clflushopt intel_pt xsaveopt xsavec xgetbv1 xsaves dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp md_clear flush_l1d cpufreq configuration: cores=2 enabledcores=2 threads=4 VPP Configuration: No workers. VPP main thread, iperf client and server are pinned to separate cores. Test Results: [11:36][ubuntu:~]$ iperf3 -s -B 200.1.1.1 -f K -A 3 --- Server listening on 5201 --- Accepted connection from 100.1.1.1, port 45188 [ 5] local 200.1.1.1 port 5201 connected to 100.1.1.1 port 45190 [ ID] Interval Transfer Bandwidth [ 5] 0.00-1.00 sec 154 KBytes 154 KBytes/sec [ 5] 1.00-2.00 sec 783 KBytes 784 KBytes/sec [ 5] 2.00-3.00 sec 782 KBytes 782 KBytes/sec [ 5] 3.00-4.00 sec 663 KBytes 663 KBytes/sec [ 5] 4.00-5.00 sec 631 KBytes 631 KBytes/sec [ 5] 5.00-6.00 sec 677 KBytes 677 KBytes/sec [ 5] 6.00-7.00 sec 693 KBytes 693 KBytes/sec [ 5] 7.00-8.00 sec 706 KBytes 706 KBytes/sec [ 5] 8.00-9.00 sec 672 KBytes 672 KBytes/sec [ 5] 9.00-10.00 sec 764 KBytes 764 KBytes/sec [ 5] 10.00-10.04 sec 21.2 KBytes 504 KBytes/sec - - - - - - - - - - - - - - - - - - - - - - - - - [ ID] Interval Transfer Bandwidth [ 5] 0.00-10.04 sec 0.00 Bytes 0.00 KBytes/sec sender [ 5] 0.00-10.04 sec 6.39 MBytes 652 KBytes/sec receiver --- Server listening on 5201 --- [11:36][ubuntu:~]$ sudo iperf3 -c 200.1.1.1 -A 2 Connecting to host 200.1.1.1, port 5201 [ 4] local 100.1.1.1 port 45190 connected to 200.1.1.1 port 5201 [ ID] Interval Transfer Bandwidth Retr Cwnd [ 4] 0.00-1.00 sec 281 KBytes 2.30 Mbits/sec 44 2.83 KBytes [ 4] 1.00-2.00 sec 807 KBytes 6.62 Mbits/sec 124 5.66 KBytes [ 4] 2.00-3.00 sec 737 KBytes 6.04 Mbits/sec 136 5.66 KBytes [ 4] 3.00-4.00 sec 720 KBytes 5.90 Mbits/sec 130 5.66 KBytes [ 4] 4.00-5.00 sec 574 KBytes 4.70 Mbits/sec 134 5.66 KBytes [ 4] 5.00-6.00 sec 720 KBytes 5.90 Mbits/sec 120 7.07 KBytes [ 4] 6.00-7.00 sec 666 KBytes 5.46 Mbits/sec 134 5.66 KBytes [ 4] 7.00-8.00 sec 741 KBytes 6.07 Mbits/sec 124 5.66 KBytes [ 4] 8.00-9.00 sec 660 KBytes 5.41 Mbits/sec 128 4.24 KBytes [ 4] 9.00-10.00 sec 740 KBytes 6.05 Mbits/sec 130 4.24 KBytes - - - - - - - - - - - - - -
Re: [vpp-dev]: Trouble shooting low bandwidth of memif interface
Looks like the image is not visible. Resending the topology diagram for reference. [image: iperf_memif.png] On Thu, Jul 30, 2020 at 11:44 AM Rajith PR via lists.fd.io wrote: > Hello Experts, > > I am trying to measure the performance of memif interface and getting a > very low bandwidth(652Kbytes/sec). I am new to performance tuning and any > help on troubleshooting the issue would be very helpful. > > The test topology i am using is as below: > > > > Basically, I have two lxc containers each hosting an instance of VPP. The > VPP instances are connected using memif. On lxc-01 i run the iperf3 client > that generates TCP traffic and on lxc-02 i run the iperf3 server. Linux > veth pairs are used for interconnecting the iperf tool with VPP. > > *Test Environment:* > > *CPU Details:* > > *-cpu > description: CPU > product: Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz > vendor: Intel Corp. > physical id: c > bus info: cpu@0 > version: Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz > serial: None > slot: U3E1 > size: 3100MHz > capacity: 3100MHz > width: 64 bits > clock: 100MHz > capabilities: x86-64 fpu fpu_exception wp vme de pse tsc msr pae > mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse > sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp constant_tsc art arch_perfmon > pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq > dtes64 monitor ds_cpl vmx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid sse4_1 > sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand > lahf_lm abm 3dnowprefetch cpuid_fault epb invpcid_single pti ssbd ibrs ibpb > stibp tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 avx2 > smep bmi2 erms invpcid mpx rdseed adx smap clflushopt intel_pt xsaveopt > xsavec xgetbv1 xsaves dtherm ida arat pln pts hwp hwp_notify hwp_act_window > hwp_epp md_clear flush_l1d cpufreq > configuration: cores=2 enabledcores=2 threads=4 > > *VPP Configuration:* > > No workers. VPP main thread, iperf client and server are pinned to > separate cores. > > *Test Results:* > > [11:36][ubuntu:~]$ iperf3 -s -B 200.1.1.1 -f K -A 3 > --- > Server listening on 5201 > --- > Accepted connection from 100.1.1.1, port 45188 > [ 5] local 200.1.1.1 port 5201 connected to 100.1.1.1 port 45190 > [ ID] Interval Transfer Bandwidth > [ 5] 0.00-1.00 sec 154 KBytes 154 KBytes/sec > [ 5] 1.00-2.00 sec 783 KBytes 784 KBytes/sec > [ 5] 2.00-3.00 sec 782 KBytes 782 KBytes/sec > [ 5] 3.00-4.00 sec 663 KBytes 663 KBytes/sec > [ 5] 4.00-5.00 sec 631 KBytes 631 KBytes/sec > [ 5] 5.00-6.00 sec 677 KBytes 677 KBytes/sec > [ 5] 6.00-7.00 sec 693 KBytes 693 KBytes/sec > [ 5] 7.00-8.00 sec 706 KBytes 706 KBytes/sec > [ 5] 8.00-9.00 sec 672 KBytes 672 KBytes/sec > [ 5] 9.00-10.00 sec 764 KBytes 764 KBytes/sec > [ 5] 10.00-10.04 sec 21.2 KBytes 504 KBytes/sec > - - - - - - - - - - - - - - - - - - - - - - - - - > [ ID] Interval Transfer Bandwidth > [ 5] 0.00-10.04 sec 0.00 Bytes 0.00 KBytes/sec > sender > [ 5] 0.00-10.04 sec 6.39 MBytes 652 KBytes/sec > receiver > --- > Server listening on 5201 > --- > > > [11:36][ubuntu:~]$ sudo iperf3 -c 200.1.1.1 -A 2 > Connecting to host 200.1.1.1, port 5201 > [ 4] local 100.1.1.1 port 45190 connected to 200.1.1.1 port 5201 > [ ID] Interval Transfer Bandwidth Retr Cwnd > [ 4] 0.00-1.00 sec 281 KBytes 2.30 Mbits/sec 44 2.83 KBytes > > [ 4] 1.00-2.00 sec 807 KBytes 6.62 Mbits/sec 124 5.66 KBytes > > [ 4] 2.00-3.00 sec 737 KBytes 6.04 Mbits/sec 136 5.66 KBytes > > [ 4] 3.00-4.00 sec 720 KBytes 5.90 Mbits/sec 130 5.66 KBytes > > [ 4] 4.00-5.00 sec 574 KBytes 4.70 Mbits/sec 134 5.66 KBytes > > [ 4] 5.00-6.00 sec 720 KBytes 5.90 Mbits/sec 120 7.07 KBytes > > [ 4] 6.00-7.00 sec 666 KBytes 5.46 Mbits/sec 134 5.66 KBytes > > [ 4] 7.00-8.00 sec 741 KBytes 6.07 Mbits/sec 124 5.66 KBytes > > [ 4] 8.00-9.00 sec 660 KBytes 5.41 Mbits/sec 128 4.24 KBytes > > [ 4] 9.00-10.00 sec 740 KBytes 6.05 Mbits/sec 130 4.24 KBytes > > - - - - - - - - - - - - - - - - - - - - - - - - - > [ ID] Interval Transfer Bandwidth Retr > [ 4] 0.00-10.00 sec 6.49 MBytes 5.44 Mbits/sec 1204 > sender > [ 4] 0.00-10.00 sec 6.39 MBytes 5.36 Mbits/sec > receiver > > iperf Done. > > Thanks, > Rajith > > -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#17116): https://lists.fd.io/g/vpp-
