Version 54 (modified by 12 years ago) (diff) | ,
---|
- Confirmation Tests Performance Measurements
-
Performance Results
- VM to VM measurements within one rack (IG-CT-1 - Access to New Site VM …
- VM to VM measurements within one rack (VMs on separate VM servers) (no …
- PC to VM measurements within one rack (IG-CT-2 - Access to New Site …
- VM to PC measurements within one rack (IG-CT-2 - Access to New Site …
- PC to PC measurements within one rack (no specific test case exist)
- One Gbits/sec Links within one rack
- VM to remote VM measurements between two non-OF sites (IG-CT-3 - …
- VM to remote OF VM measurements between multiple OpenFlow sites …
Confirmation Tests Performance Measurements
This page captures all performance measurements results captured during InstaGENI New Site Confirmation tests. For test details see the InstaGENI Confirmation Tests page. For test status see the InstaGENI New Site Confirmation Tests Status page.
Test Methodology
Test Characteristics:
- All VMs at one site are on the same VM server, unless otherwise stated.
- All results are captured for default bandwidth allocation (100Mb/s shaped), unless otherwise stated.
- All results are captured for default OS (Fedora 15 STD).
- All results collected between 2013-03-01 through 2013-03-07, unless otherwise stated.
Iperf TCP measurements captured in this page have the following assumptions:
- All results are for a 60 second test run.
- for OpenFlow scenarios, traffic is exchanged to establish data flows before measurement test runs are executed.
- iperf '1 client' scenario command: 'iperf -c dest_host -t 60'
- iperf '5 clients' scenario command: 'iperf -c dest_host -t 60 -P 5'
- iperf '10 clients' scenario command: 'iperf -c dest_host -t 60 -P 10'
Iperf UDP measurements captured in this page have the following assumptions:
- Each test run is 10 seconds.
- UDP measurements within a rack requested a bandwidth of 1Gbits/sec for test run
- UDP measurements between racks requested a bandwidth of 100 Mbits/sec, if results showed that 100Mb/s was possible, then increased to 1Gb/s.
Performance Results
VM to VM measurements within one rack (IG-CT-1 - Access to New Site VM resources)
End points | One Client(tcp) | 5 Clients(tcp) | 10 Clients(tcp) | 1 client(udp) | Ping rtt min/avg/max/mdev |
GPO | 96.9 Mbits/sec | 97.8 Mbits/sec | 99.3 Mbits/sec | 95.7 Mbits/sec | 0.026/0.030/0.040/0.004 ms |
Utah | 96.9 Mbits/sec | 97.7 Mbits/sec | 99.1 Mbits/sec | 96.7 Mbits/sec | 0.028/0.031/0.053/0.008 ms |
Northwestern | 96.9 Mbits/sec | 97.9 Mbits/sec | 99.2 Mbits/sec | 97.4 Mbits/sec | 0.025/0.033/0.248/0.028 ms |
Kentucky | 96.9 Mbits/sec | 98.0 Mbits/sec | 98.9 Mbits/sec | 95.2 Mbits/sec | 0.025/0.082/3.229/0.409 ms |
GATech | 96.9 Mbits/sec | 98.0 Mbits/sec | 99.2 Mbits/sec | 98.0 Mbits/sec | 0.026/0.030/0.047/0.005 ms |
VM to VM measurements within one rack (VMs on separate VM servers) (no specific test case exists)
Even though not part of planned tests, ran an experiment with 2 VMs each on a separate VM server to capture performance between VM nodes on different servers within the same rack.
End points | One Client(tcp) | 5 Clients(tcp) | 10 Clients(tcp) | 1 client(udp) | Ping rtt min/avg/max/mdev |
GPO | 101 Mbits/sec | 101 Mbits/sec | 101 Mbits/sec | 96.7 Mbits/sec | 0.134/0.165/0.184/0.014 ms |
PC to VM measurements within one rack (IG-CT-2 - Access to New Site Raw PC and VM resources)
End points | One Client(tcp) | 5 Clients(tcp) | 10 Clients(tcp) | 1 client(udp) | Ping rtt min/avg/max/mdev |
GPO | 101 Mbits/sec | 101 Mbits/sec | 101 Mbits/sec | 96.8 Mbits/sec | 0.123/0.132/0.161/0.015 ms |
Northwestern* | 101 Mbits/sec | 101 Mbits/sec | 101 Mbits/sec | ---- Mbits/sec | 0.175/0.189/0.238/0.014 ms |
Utah | 101 Mbits/sec | 101 Mbits/sec | 101 Mbits/sec | 96.8 Mbits/sec | 0.118/0.133/0.170/0.015 ms |
Kentucky | 101 Mbits/sec | 101 Mbits/sec | 101 Mbits/sec | 96.8 Mbits/sec | 0.158/0.188/0.224/0.013 ms |
GATech | 101 Mbits/sec | 101 Mbits/sec | 100 Mbits/sec | 97.9 Mbits/sec |
(*) need to re-execute when raw-pc is available - results collected on 2013-02-20
VM to PC measurements within one rack (IG-CT-2 - Access to New Site Raw PC and VM resources)
End points | One Client(tcp) | 5 Clients(tcp) | 10 Clients(tcp) | 1 client(udp) | Ping rtt min/avg/max/mdev |
GPO | 101 Mbits/sec | 101 Mbits/sec | 102 Mbits/sec | 98.0 Mbits/sec | 0.134/0.148/0.168/0.016 ms |
Northwestern* | ----- | ----- | ----- | ----- | ----- |
Utah | 101 Mbits/sec | 101 Mbits/sec | 102 Mbits/sec | 98.0 Mbits/sec | 0.104/0.139/0.146/0.017 ms |
Kentucky | 101 Mbits/sec | 101 Mbits/sec | 102 Mbits/sec | 96.2 Mbits/sec | 0.146/0.157/0.174/0.006 ms |
(*) need to capture when raw-pc is available - no results have been collected for this in past test runs
PC to PC measurements within one rack (no specific test case exist)
End points | One Client(tcp) | 5 Clients(tcp) | 10 Clients(tcp) | 1 client(udp) | Ping rtt min/avg/max/mdev |
GPO | 941 Mbits/sec | 942 Mbits/sec | 942 Mbits/sec | --- Mbits/sec | 0.139/0.183/0.229/0.015 ms |
Northwestern (*) | 942 Mbits/sec | 942 Mbits/sec | 942 Mbits/sec | --- Mbits/sec | 0.123/0.134/0.149/0.010 ms |
GATech | 941 Mbits/sec | 942 Mbits/sec | 942 Mbits/sec | 101 Mbits/sec | 0.108/0.134/0.145/0.016 ms |
(*) Collected on 2013-02-20 - need to re-capture when raw-pcs are available after demo.
One Gbits/sec Links within one rack
For this test case, each link was requested with a speed of 1 Gb/s. All tests were conducted in the GPO Rack on 2013-03-01.
End points | One Client(tcp) | 5 Clients(tcp) | 10 Clients(tcp) | 1 client(udp) | Ping rtt min/avg/max/mdev |
VM to VM (1 VM server) | 962 Mbits/sec | 962 Mbits/sec | 964 Mbits/sec | 808 Mbits/sec | 0.022/0.026/0.036/0.004 ms |
VM (server1) to VM (server2) | 929 Mbits/sec | 935 Mbits/sec | 935 Mbits/sec | 809 Mbits/sec | 0.108/0.145/0.165/0.010 ms |
PC to VM | 939 Mbits/sec | 940 Mbits/sec | 940 Mbits/sec | 806 Mbits/sec | 0.123/0.175/0.215/0.017 ms |
VM to PC | 939 Mbits/sec | 938 Mbits/sec | 936 Mbits/sec | 808 Mbits/sec | 0.138/0.144/0.160/0.016 ms |
PC to PC | 941 Mbits/sec | 942 Mbits/sec | 942 Mbits/sec | ---- Mbits/sec | 0.136/0.174/0.192/0.012 ms |
VM to remote VM measurements between two non-OF sites (IG-CT-3 - Multiple Sites experiment)
End points | One Client(tcp) | 5 Clients(tcp) | 10 Clients(tcp) | 1 client(udp) | Ping rtt min/avg/max/mdev |
Utah IG to GPO IG | 5.12 Mbits/sec | 22.5 Mbits/sec | 27.2 Mbits/sec | 74.6 Mbits/sec | 59.270/62.294/71.982/3.371 ms |
GPO IG to Utah IG | 19.5 Mbits/sec | 39.8 Mbits/sec | 46.2 Mbits/sec | 69.3 Mbits/sec | 59.187/62.125/71.587/3.282 ms |
GPO IG to Northwestern | 27.5 Mbits/sec | 51.3 Mbits/sec | 64.5 Mbits/sec | 74.2 Mbits/sec | 24.136/27.096/36.095/2.587 ms |
Northwestern to GPO IG | 10.2 Mbits/sec | 24.7 Mbits/sec | 30.5 Mbits/sec | 67.5 Mbits/sec | 24.127/29.711/56.317/5.606 ms |
Kentucky to GPO | 41.2 Mbits/sec | 54.6 Mbits/sec | 56.6 Mbits/sec | 72.7 Mbits/sec | 40.820/41.786/50.091/1.570 ms |
GPO to Kentucky | 53.4 Mbits/sec | 65.7 Mbits/sec | 64.7 Mbits/sec | 75.2 Mbits/sec | 40.736/41.645/51.870/2.115 ms |
VM to remote OF VM measurements between multiple OpenFlow sites (IG-CT-4 - Multiple sites OpenFlow experiment and interoperability)
All OpenFlow combination use default Operating System and default link bandwidth.
End points | One Client(tcp) | 5 Clients(tcp) | 10 Clients(tcp) | 1 client(udp) | Ping rtt min/avg/max/mdev |
GPO IG to GPO IG | 926 Mbits/sec | 932 Mbits/sec | 936 Mbits/sec | 810 Mbits/sec | 0.066/0.181/6.499/0.822 ms |
Northwestern to Northwestern | 939 Mbits/sec | 938 Mbits/sec | 935 Mbits/sec | 809 Mbits/sec | 0.066/1.044/58.357/7.461 ms |
Utah IG to Utah IG | 934 Mbits/sec | 933 Mbits/sec | 934 Mbits/sec | 812 Mbits/sec | 0.069/2.123/122.830/15.714 ms |
UKY IG to UKY IG | 930 Mbits/sec | 934 Mbits/sec | 935 Mbits/sec | 808 Mbits/sec | 0.069/1.567/89.650/11.467 ms |
GPO IG to Northwestern | 13.1 Mbits/sec | 55.8 Mbits/sec | 110 Mbits/sec | 101 Mbits/sec | 24.267/30.973/422.718/51.000 ms |
Northwestern to GPO IG | 890 Mbits/sec | 864 Mbits/sec | 832 Mbits/sec | 808 Mbits/sec | 24.320/31.271/439.747/53.179 ms |
GPO IG to Utah IG (*) | 28.1 Mbits/se | 34.6 Mbits/sec | 44.1 Mbits/sec | 95.4 Mbits/sec | 270.577/420.450/3926.879/614.078 ms |
GPO IG to Utah IG () | 41.3 Mbits/sec | 56.4 Mbits/sec | 68.3 Mbits/sec | 95.5 Mbits/sec | 85.995/107.969/1232.550/147.996 ms |
Utah IG to GPO IG (*) | 32.4 Mbits/sec | 67.6 Mbits/sec | 67.7 Mbits/sec | 95.7 Mbits/sec | 270.624/384.656/3385.019/496.738 ms |
Utah IG to GPO IG () | 77.7 Mbits/sec | 87.1 Mbits/sec | 91.2 Mbits/sec | 95.5 Mbits/sec | 85.990/104.017/1097.500/129.613 ms |
UKY IG to GPO IG | 66.2 Mbits/sec | 253 Mbits/sec | 249 Mbits/sec | 809 Mbits/sec | 174.400/235.158/2376.246/320.207 ms |
GPO IG to UKY IG | 121 Mbits/sec | 605 Mbits/sec | 587 Mbits/sec | 809 Mbits/sec | 174.405/232.141/2286.578/305.249 ms |
GPO IG to Utah PG | 23.8 Mbits/sec | 43.6 Mbits/sec | 45.9 Mbits/sec | 932 Kbits/sec | 270.547/379.351/3300.089/482.218 ms |
Northwestern to Utah PG | 34.1 Mbits/sec | 38.0 Mbits/sec | 49.0 Mbits/sec | 95.0 Mbits/sec | 246.707/343.387/3153.966/457.641 ms |
Utah IG to Utah PG | 933 Mbits/sec | 928 Mbits/sec | 936 Mbits/sec | 808 Mbits/sec | 0.157/2.285/123.481/15.778 ms |
UKY IG to Utah PG | 47.4 Mbits/sec | 56.9 Mbits/sec | 70.9 Mbits/sec | 95.5 Mbits/sec | 121.731/148.954/1430.618/171.832 ms |
GPO IG to GPO EG | 854 Mbits/sec | 865 Mbits/sec | 861 Mbits/sec | 808 Mbits/sec | 0.155/1.294/48.102/6.194 ms |
Utah IG to GPO EG | 3.79 Mbits/sec | 18.6 Mbits/sec | 35.2 Mbits/sec | 95.7 Mbits/sec | 270.673/391.340/3453.525/510.477 ms |
Utah PG to GPO EG | 3.82 Mbits/sec | 18.2 Mbits/sec | 35.1 Mbits/sec | 96.4 Mbits/sec | 270.619/381.588/3371.182/493.036 ms |
Northwestern to RENCI EG | 6.93 Mbits/sec | 34.3 Mbits/sec | 68.6 Mbits/sec | 806 Mbits/sec | 147.632/214.587/2465.440/341.732 ms |
UKY IG to RENCI EG | 4.00 Mbits/sec | 22.7 Mbits/sec | 46.0 Mbits/sec | 102 Mbits/sec | 174.456/246.732/2609.205/363.068 ms |
Northwestern to WAPG Rutgers | 130 Mbits/sec | 547 Mbits/sec | 850 Mbits/sec | 805 Mbits/sec | 155.140/227.661/2577.212/361.250 ms |
GPO IG to WAPG Rutgers | 112 Mbits/sec | 492 Mbits/sec | 815 Mbits/sec | 101 Mbits/sec | 179.018/243.752/2410.066/330.414 ms |
Utah IG to WAPG Rutgers | 74.1 Mbits/sec | 84.1 Mbits/sec | 89.7 Mbits/sec | 101 Mbits/sec | 100.676/122.137/1227.486/145.294 ms |
UKY IG to WAPG Rutgers | 672 Mbits/sec | 780 Mbits/sec | 742 Mbits/sec | 736 Mbits/sec | 30.161/39.489/586.414/71.203 ms |
Utah PG to Utah IG | 928 Mbits/sec | 932 Mbits/sec | 933 Mbits/sec | 806 Mbits/sec | 0.167/22.625/134.880/50.201 ms |
Utah PG to GPO IG (*) | 34.1 Mbits/sec | 64.1 Mbits/sec | 69.8 Mbits/sec | 96.0 Mbits/sec | 270.547/352.256/2868.660/397.82 ms |
(*) Longest Path: GPO IG ->NLR CHIC->NLR DENV->NLR SEAT->NLR SUN->NLR ATLA->I2 ATLA->I2 WASH->I2 NEWY->I2 LOSA->UTAH IG
() Shortest Path: GPO IG ->I2 NEWY->I2 LOSA-> UTAH IG