Problems with link to/from BNLAdnan Iqbal. Page created: Jan 05, 2006Central Computer Access | Computer Networking | Network Group | More case studies |
Monitoring Node |
Alert/Time/Drop |
|
IEPM-BW.BNL.GOV |
IEPM-BW.SLAC.STANFORD.EDU |
1) 12/30/2005 01:32:34, %drop = 40.9 |
NODE1.CACR.CALTECH.EDU |
1) 12/28/2005 23:10:15, %drop = 37.5 2) 12/29/2005 09:41:17, %drop = 46.7 |
|
NODE1.PD.INFN.IT |
1) 12/28/2005 21:28:54, %drop = 45.6 2) 12/29/2005 05:32:21, %drop = 51.4 3) 12/29/2005 22:33:49, %drop = 49.4 |
|
IEPM-BW.CERN.CH |
1) 12/29/2005 04:21:48, %%drop = 48.5 2) 12/29/2005 06:46:51, %%drop = 48.5 |
|
NODE1.CACR.CALTECH.EDU |
1) 12/29/2005 09:41:17, %%drop = 46.7 |
|
NODE1.DL.AC.UK |
2) 12/29/2005 01:52:17, %%drop = 34.4 |
|
Total Nodes = 6 |
Number of Alerts = 10 Earliest Detection 12/28/2005 21:28:54 Latest Detection 12/30/2005 01:32:34 |
We received number of e-mails from iepm-bw.bnl.gov about alerts for different monitored nodes between the time period 12/28/2005 21:28:54 - 12/30/2005 01:32:34 These events were detected by Path chirp analysis tool using Plateau algorithm.
At first we tried to find out any route change in the specific time period. We came to know that there were two nodes who experienced a route change but that route change was very small and cannot be associated with any adverse effect.
Our second step was to figure out what is the effect on other nodes monitored by iepm-bw.bnl.gov. We observed that majority of monitored sites experienced decrease in performance, some of them increase in performance and very few did not experience any thing.
NODE |
COMMENT |
iepmbw.cacr.caltech.edu |
-ve, no route change, trigger unstable, 1 alert |
node1.cacr.caltech.edu |
-ve, no route change, trigger buf looking stablizing, 2 alert |
node1.sdsc.edu |
-ve, no route change, trg buf getting worse, 0 alert |
node1.stanford.edu |
+ve, no route change getting better |
iepmbw.slac.stanford.edu |
-ve, no route change, trg buf unstable, 1 alert |
node2.slac.stanford.edu |
-ve, no route change, trg buf getting worse, 0 alert |
node1.indiana.edu |
+ve, no route change getting better |
node1.dl.ac.uk |
-ve, no route change, trg buf unstable, 1 alert |
node1.mcs.anl.gov |
-ve, no route change, trg buf unstable, 0 alert |
node1.desy.de |
+ve, no route change getting better |
node1.fzk.de |
+ve, no route change getting better |
iepmbw.cern.ch |
-ve, no route change, trg buf stablizing, 2 alert |
node1.bnp.nsk.su |
no change |
node1.pd.infn.it |
-ve, no route change, trg buf unstable, 3 alert |
node1.utdallas.edu |
+ve, 2 minor route change getting better |
node2.nslabs.ufl.edu |
-ve, no route change, trg buf unstable, 0 alert |
node1.utoranto.ca |
-ve, 2 route change (minor) , trg buf unstable, 0 alert |
Now we tried to find out does any network experience bad behavior. for this purpose, we separated all monitored nodes with respect to their trace route similarity and compared the reported performance among the group of similar trace routes. we came up with the result that there is no correlation between the network behavior and the node behavior. Following sheet describes it in detail. Similar color shows the same path up to that hop.
NODE | COMMENT | H1 | H2 | H3 | H4 | H5 | H6 | H7 | H8 | H9 | H10 | H11 | H12 | H13 | H14 | H15 | H16 | H17 |
iepmbw.cacr.caltech.edu | -ve, no route change, trigger unstable, 1 alert | anubis.s80.bnl.gov 130.199.80.124 0.867 ms | shu.v400.bnl.gov 130.199.136.9 0.349 ms | amon 130.199.3.24 0.694 ms | esbnl-bnl.es.net 198.124.216.113 0.623 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.342 ms | chicr1-oc192-aoacr1.es.net 134.55.209.57 22.347 ms | snvcr1-oc192-chicr1.es.net 134.55.209.53 70.417 ms | snv2sdn1-snvcr1.es.net 134.55.217.10 70.443 ms | cenichpr-1-lo-jmb-704.snvaca.pacificwave.net 207.231.244.1 70.600 ms | lax-hpr--svl-hpr-10ge.cenic.net 137.164.25.12 78.258 ms | lax-hpr.losnettos-hpr.cenic.net 137.164.27.246 78.444 ms | Booth-RSM.ilan.caltech.edu 131.215.254.253 78.604 ms | CACR-ITS-BMR.caltech.edu 131.215.5.147 78.550 ms | iepm-bw.cacr.caltech.edu 131.215.xxx.xxx 78.309 ms | |||
node1.cacr.caltech.edu | -ve, no route change, trigger buf looking stablizing, 2 alert | anubis.s80.bnl.gov 130.199.80.124 0.326 ms | shu.v400.bnl.gov 130.199.136.9 0.401 ms | amon 130.199.3.24 0.781 ms | esbnl-bnl.es.net 198.124.216.113 0.739 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.444 ms | chicr1-oc192-aoacr1.es.net 134.55.209.57 27.952 ms | snvcr1-oc192-chicr1.es.net 134.55.209.53 70.418 ms | snv2sdn1-snvcr1.es.net 134.55.217.10 70.469 ms | cenichpr-1-lo-jmb-704.snvaca.pacificwave.net 207.231.244.1 70.941 ms | lax-hpr--svl-hpr-10ge.cenic.net 137.164.25.12 78.379 ms | lax-hpr.losnettos-hpr.cenic.net 137.164.27.246 78.558 ms | Booth-RSM.ilan.caltech.edu 131.215.254.253 78.530 ms | CACR-ITS-BMR.caltech.edu 131.215.5.147 78.505 ms | node1.cacr.caltech.edu 131.215.xxx.xxx 78.211 ms | |||
node1.sdsc.edu | -ve, no route change, trg buf getting worse, 0 alert | anubis.s80.bnl.gov 130.199.80.124 0.241 ms | shu.v400.bnl.gov 130.199.136.9 0.310 ms | amon 130.199.3.24 0.654 ms | esbnl-bnl.es.net 198.124.216.113 0.566 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.321 ms | chicr1-oc192-aoacr1.es.net 134.55.209.57 22.353 ms | snvcr1-oc192-chicr1.es.net 134.55.209.53 70.435 ms | snv2sdn1-snvcr1.es.net 134.55.217.10 70.521 ms | cenichpr-1-lo-jmb-704.snvaca.pacificwave.net 207.231.244.1 70.697 ms | lax-hpr--svl-hpr-10ge.cenic.net 137.164.25.12 78.301 ms | riv-hpr--lax-hpr-10ge.cenic.net 137.164.25.5 84.360 ms | hpr-sdsc-sdsc2--riv-hpr-ge.cenic.net 137.164.27.54 84.145 ms | thunder.sdsc.edu 132.249.30.5 84.076 ms | node1.sdsc.edu 132.249.xxx.xxx 84.136 ms | |||
node1.stanford.edu | +ve, no route change getting better | anubis.s80.bnl.gov 130.199.80.124 0.317 ms | shu.v400.bnl.gov 130.199.136.9 0.284 ms | amon 130.199.3.24 0.656 ms | esbnl-bnl.es.net 198.124.216.113 0.568 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.298 ms | chicr1-oc192-aoacr1.es.net 134.55.209.57 22.328 ms | snvcr1-oc192-chicr1.es.net 134.55.209.53 70.428 ms | snv2sdn1-snvcr1.es.net 134.55.217.10 70.522 ms | cenichpr-1-lo-jmb-704.snvaca.pacificwave.net 207.231.244.1 70.667 ms | hpr-stan-ge--svl-hpr.cenic.net 137.164.27.162 70.791 ms | bbr2-rtr.Stanford.EDU 171.64.1.133 71.226 ms | rtf-rtr.Stanford.EDU 171.64.1.162 71.815 ms | noc6.Stanford.EDU 171.64.xxx.xxx 70.940 ms | ||||
iepmbw.slac.stanford.edu | -ve, no route change, trg buf unstable, 1 alert | anubis.s80.bnl.gov 130.199.80.124 0.289 ms | shu.v400.bnl.gov 130.199.136.9 0.361 ms | amon 130.199.3.24 0.777 ms | esbnl-bnl.es.net 198.124.216.113 0.842 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.476 ms | chicr1-oc192-aoacr1.es.net 134.55.209.57 22.522 ms | snvcr1-oc192-chicr1.es.net 134.55.209.53 70.558 ms | snv1mr1-snvcr1.es.net 134.55.218.22 70.668 ms | snv2mr1-snv1mr1.es.net 134.55.217.6 79.437 ms | slacmr1-snv2mr1.es.net 134.55.217.1 71.003 ms | slacrt4-slacmr1.es.net 134.55.209.94 71.141 ms | rtr-dmz1-vlan400.slac.stanford.edu 192.68.191.149 71.119 ms | iepm-bw.slac.stanford.edu 134.79.xxx.xxx 71.115 ms | ||||
node2.slac.stanford.edu | -ve, no route change, trg buf getting worse, 0 alert | anubis.s80.bnl.gov 130.199.80.124 0.202 ms | shu.v400.bnl.gov 130.199.136.9 0.279 ms | amon 130.199.3.24 0.736 ms | esbnl-bnl.es.net 198.124.216.113 0.779 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.449 ms | chicr1-oc192-aoacr1.es.net 134.55.209.57 22.494 ms | snvcr1-oc192-chicr1.es.net 134.55.209.53 70.541 ms | snv1mr1-snvcr1.es.net 134.55.218.22 70.564 ms | snv2mr1-snv1mr1.es.net 134.55.217.6 79.117 ms | slacmr1-snv2mr1.es.net 134.55.217.1 70.961 ms | slacrt4-slacmr1.es.net 134.55.209.94 71.108 ms | rtr-dmz1-vlan400.slac.stanford.edu 192.68.191.149 71.078 ms | node2.slac.stanford.edu 134.79.xxx.xxx 71.053 ms | ||||
node1.indiana.edu | +ve, no route change getting better | anubis.s80.bnl.gov 130.199.80.124 0.535 ms | shu.v400.bnl.gov 130.199.136.9 0.393 ms | amon 130.199.3.24 0.657 ms | esbnl-bnl.es.net 198.124.216.113 0.626 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.432 ms | chicr1-oc192-aoacr1.es.net 134.55.209.57 22.744 ms | chi-esnet.abilene.iu.edu 198.125.140.54 22.411 ms | iplsng-chinng.abilene.ucaid.edu 198.32.8.77 31.373 ms | ul-abilene.indiana.gigapop.net 192.12.206.250 26.407 ms | 149.165.254.230 149.165.254.230 26.491 ms | icr2-ibs1-pb.noc.iu.edu 149.166.2.32 26.418 ms | node1.indiana.edu 134.68.xxx.xxx 26.176 ms | |||||
node1.dl.ac.uk | -ve, no route change, trg buf unstable, 1 alert | anubis.s80.bnl.gov 130.199.80.124 0.267 ms | shu.v400.bnl.gov 130.199.136.9 0.399 ms | amon 130.199.3.24 0.744 ms | esbnl-bnl.es.net 198.124.216.113 0.587 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.345 ms | esnet.ny1.ny.geant.net 62.40.105.25 2.406 ms | ny.uk1.uk.geant.net 62.40.96.170 71.155 ms | po2-0-0.gn2-gw1.ja.net 62.40.124.198 71.104 ms | po1-1.lond-scr3.ja.net 146.97.35.97 71.066 ms | po0-0.read-scr.ja.net 146.97.33.38 72.266 ms | po3-0.warr-scr.ja.net 146.97.33.54 75.939 ms | po1-0.manchester-bar.ja.net 146.97.35.166 76.308 ms | gw-nnw.core.netnw.net.uk 146.97.40.202 76.505 ms | gw-liv.core.netnw.net.uk 194.66.25.10 77.320 ms | gw-fw.dl.ac.uk 193.63.74.233 77.758 ms | alan3.dl.ac.uk 193.63.74.129 77.961 ms | node1.dl.ac.uk 193.62.xxx.xxx 78.096 ms |
node1.mcs.anl.gov | -ve, no route change, trg buf unstable, 0 alert | anubis.s80.bnl.gov 130.199.80.124 0.267 ms | shu.v400.bnl.gov 130.199.136.9 0.399 ms | amon 130.199.3.24 0.744 ms | esbnl-bnl.es.net 198.124.216.113 0.587 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.345 ms | esnet.ny1.ny.geant.net 62.40.105.25 2.406 ms | ny.uk1.uk.geant.net 62.40.96.170 71.155 ms | po2-0-0.gn2-gw1.ja.net 62.40.124.198 71.104 ms | po1-1.lond-scr3.ja.net 146.97.35.97 71.066 ms | po0-0.read-scr.ja.net 146.97.33.38 72.266 ms | po3-0.warr-scr.ja.net 146.97.33.54 75.939 ms | po1-0.manchester-bar.ja.net 146.97.35.166 76.308 ms | gw-nnw.core.netnw.net.uk 146.97.40.202 76.505 ms | gw-liv.core.netnw.net.uk 194.66.25.10 77.320 ms | gw-fw.dl.ac.uk 193.63.74.233 77.758 ms | alan3.dl.ac.uk 193.63.74.129 77.961 ms | node1.dl.ac.uk 193.62.xxx.xxx 78.096 ms |
node1.desy.de | +ve, no route change getting better | anubis.s80.bnl.gov 130.199.80.124 0.251 ms | shu.v400.bnl.gov 130.199.136.9 0.347 ms | amon 130.199.3.24 0.718 ms | esbnl-bnl.es.net 198.124.216.113 0.739 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.340 ms | esnet.ny1.ny.geant.net 62.40.105.25 2.575 ms | ny.uk1.uk.geant.net 62.40.96.170 71.118 ms | uk.fr1.fr.geant.net 62.40.96.89 78.028 ms | fr.ch1.ch.geant.net 62.40.96.29 86.197 ms | so-7-2-0.rt1.fra.de.geant2.net 62.40.112.22 94.267 ms | dfn-gw.rt1.fra.de.geant2.net 62.40.124.34 94.354 ms | cr-berlin1-po1-0.x-win.dfn.de 188.1.18.53 111.416 ms | cr-hamburg1-po10-0.x-win.dfn.de 188.1.18.110 111.574 ms | 188.1.47.42 188.1.47.42 111.456 ms | node1.desy.de 131.169.xxx.xxx 111.445 ms | ||
node1.fzk.de | +ve, no route change getting better | anubis.s80.bnl.gov 130.199.80.124 0.229 ms | shu.v400.bnl.gov 130.199.136.9 0.356 ms | amon 130.199.3.24 94.310 ms | esbnl-bnl.es.net 198.124.216.113 0.576 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.340 ms | esnet.ny1.ny.geant.net 62.40.105.25 2.467 ms | ny.uk1.uk.geant.net 62.40.96.170 71.066 ms | uk.fr1.fr.geant.net 62.40.96.89 78.006 ms | fr.ch1.ch.geant.net 62.40.96.29 86.145 ms | so-7-2-0.rt1.fra.de.geant2.net 62.40.112.22 94.341 ms | dfn-gw.rt1.fra.de.geant2.net 62.40.124.34 94.235 ms | ar-karlsruhe1-po6-0.x-win.dfn.de 188.1.18.94 97.230 ms | kr-fzk1.g-win.dfn.de 188.1.38.222 97.193 ms | node1.fzk.de 192.108.xxx.xxx 97.395 ms | node1.fzk.de 192.108.xxx.xxx 97.167 ms | ||
iepmbw.cern.ch | -ve, no route change, trg buf stablizing, 2 alert | anubis.s80.bnl.gov 130.199.80.124 0.207 ms | shu.v400.bnl.gov 130.199.136.9 0.258 ms | amon 130.199.3.24 0.617 ms | esbnl-bnl.es.net 198.124.216.113 0.609 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.259 ms | esnet.ny1.ny.geant.net 62.40.105.25 5.777 ms | ny.uk1.uk.geant.net 62.40.96.170 71.077 ms | uk.fr1.fr.geant.net 62.40.96.89 77.937 ms | fr.ch1.ch.geant.net 62.40.96.29 86.139 ms | swiCE2-10GE-1-1.switch.ch 62.40.124.22 91.433 ms | e513-e-rci76-1-swice2.cern.ch 192.65.184.222 114.837 ms | e513-e-rci65-3-vlan2.cern.ch 192.65.192.6 114.819 ms | iepm-bw.cern.ch 192.91.xxx.xxx 114.570 ms | ||||
node1.bnp.nsk.su | no change | anubis.s80.bnl.gov 130.199.80.124 0.275 ms | shu.v400.bnl.gov 130.199.136.9 0.357 ms | amon 130.199.3.24 0.652 ms | esbnl-bnl.es.net 198.124.216.113 0.566 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.272 ms | esnet.ny1.ny.geant.net 62.40.105.25 2.452 ms | ny.uk1.uk.geant.net 62.40.96.170 71.207 ms | uk.se1.se.geant.net 62.40.96.125 106.129 ms | rbnet-gw.se1.se.geant.net 62.40.103.42 153.827 ms | MSK-M9-RBNet-7.RBNet.ru 195.209.14.185 153.950 ms | NSK-RBNet-3.RBNet.ru 195.209.14.74 201.869 ms | 217.79.60.46 217.79.60.46 204.263 ms | warrior.inp.nsk.su 193.124.164.42 208.407 ms | zhmur.inp.nsk.su 193.124.164.17 211.262 ms | 193.124.164.5 193.124.164.5 215.733 ms | nko-315-1.inp.nsk.su 193.124.165.54 211.630 ms | node1.binp.nsk.su 193.124.xxx.xxx 210.823 ms |
node1.pd.infn.it | -ve, no route change, trg buf unstable, 3 alert | anubis.s80.bnl.gov 130.199.80.124 0.287 ms | shu.v400.bnl.gov 130.199.136.9 0.307 ms | amon 130.199.3.24 0.804 ms | esbnl-bnl.es.net 198.124.216.113 0.806 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.531 ms | esnet.ny1.ny.geant.net 62.40.105.25 2.641 ms | ny.at1.at.geant.net 62.40.96.110 106.432 ms | at.de2.de.geant.net 62.40.96.58 118.971 ms | de.it1.it.geant.net 62.40.96.62 128.013 ms | garr-gw.it1.it.geant.net 62.40.103.190 128.141 ms | rt-mi1-rt-pd1.pd1.garr.net 193.206.134.46 131.287 ms | infnpd-rtg.pd.garr.net 193.206.142.18 131.446 ms | node1.pd.infn.it 193.206.xxx.xxx 142.159 ms | ||||
node1.utdallas.edu | +ve, 2 minor route change getting better | anubis.s80.bnl.gov 130.199.80.124 0.341 ms | shu.v400.bnl.gov 130.199.136.9 0.322 ms | amon 130.199.3.24 0.790 ms | esbnl-bnl.es.net 198.124.216.113 0.636 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.393 ms | dccr1-oc48-aoacr1.es.net 134.55.209.62 6.646 ms | atlcr1-oc48-dccr1.es.net 134.55.209.66 80.640 ms | abilene-atlcr1.es.net 198.124.216.142 22.238 ms | hstnng-atlang.abilene.ucaid.edu 198.32.8.33 41.774 ms | ntg-gw1-so-3-0-0.northtexasgigapop.org 206.223.141.69 47.083 ms | utd-ntg-gw1.northtexasgigapop.org 206.223.141.74 47.006 ms | 129.110.5.20 129.110.5.20 48.472 ms | node1.utdallas.edu 129.110.xxx.xxx 47.510 ms | ||||
node2.nslabs.ufl.edu | -ve, no route change, trg buf unstable, 0 alert | anubis.s80.bnl.gov 130.199.80.124 0.300 ms | shu.v400.bnl.gov 130.199.136.9 0.303 ms | amon 130.199.3.24 0.718 ms | esbnl-bnl.es.net 198.124.216.113 0.666 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.407 ms | dccr1-oc48-aoacr1.es.net 134.55.209.62 6.530 ms | atlcr1-oc48-dccr1.es.net 134.55.209.66 22.176 ms | abilene-atlcr1.es.net 198.124.216.142 22.203 ms | jax-flrcore-7609-1-te23-1800.net.flrnet.org 198.32.155.193 28.649 ms | ssrb230a-ewan-msfc-1-v1805-1.ns.ufl.edu 198.32.155.222 33.839 ms | ssrb6c-nexus-msfc-1-v30-1.ns.ufl.edu 128.227.236.81 33.723 ms | ssrb230a-core-msfc-1-v20-1.ns.ufl.edu 128.227.236.2 33.804 ms | ssrb201-nslabs-rsm-1-v222-1.ns.ufl.edu 128.227.129.158 34.335 ms | node2.nslabs.ufl.edu 128.227.xxx.xxx 33.593 ms | |||
node1.utoranto.ca | -ve, 2 route change (minor) , trg buf unstable, 0 alert | anubis.s80.bnl.gov 130.199.80.124 0.516 ms | shu.v400.bnl.gov 130.199.136.9 0.308 ms | amon 130.199.3.24 0.728 ms | esbnl-bnl.es.net 198.124.216.113 0.743 ms | aoacr1-oc48-bnl.es.net 134.55.209.129 2.509 ms | chislsdn1-chicr1.es.net 134.55.207.34 22.626 ms | c4-tor01.canet4.net 198.125.140.142 33.012 ms | c4-orano-tor.canet4.net 205.189.32.213 32.871 ms | DIST1-TORO-GE1-1.IP.orion.on.ca 66.97.16.122 33.236 ms | GTANET-ORION-RNE.DIST1-TORO.IP.orion.on.ca 66.97.23.58 33.300 ms | utoronto-ut-hub-if.gtanet.ca 205.211.94.130 33.393 ms | esc-gateway2.gw.utoronto.ca 128.100.200.102 34.166 ms | mcl-gpb.gw.utoronto.ca 128.100.96.10 33.912 ms | bigmac-fw.physics.utoronto.ca 128.100.190.2 34.854 ms | node1.utoronto.ca 128.100.xxx.xxx 34.349 ms |
Our next attempt was to analyze the behavior of iepm-bw.bnl.gov the statistics gave following results at time 12/30/05 about 17:00:00, but it was the time when the storm had passed.
Host at BNL = IEPM-BW.BNL.GOV
/usr/binb/top shows it is >98% idle, 70MB free memory, swap space barely used
/sbin/ifconfig showed no interface errors
The host up for 2 days:
21:45:27 up 2 days, 9:48, 1 user, load average: 0.01, 0.03, 0.06
Possibly after it rebooted 2 days ago something was not set right.
The max TCP window/buffer OK, and anyway this would have only affect the iperf measurements not Pathchirp.
cat /proc/sys/net/core/wmem_max = 1048576
;cat /proc/sys/net/core/rmem_max = 1048576
;cat /proc/sys/net/core/rmem_default = 1048576
;cat /proc/sys/net/core/wmem_default = 1048576
;cat /proc/sys/net/ipv4/tcp_rmem = 262144 1048576 8388608
;cat /proc/sys/net/ipv4/tcp_wmem = 262144 1048576 8388608
The ping RTT to SLAC = 70ms, so if we say the bottleneck bandwidth should be about 500Mbits/s then the Bandwidth Delay Product is ~ 4MB so doubling that we have ~ 8MB.
Iperf from BNL to SLAC (iepm-resp) 20 secs measurement unless otherwise mentioned:
Streams Window Achievable throughput
64 256KB 552Mbits/s
64 128KB 496Mbits/s
32 256KB 522Mbits/s
16 128KB 261Mbits/s
16 512KB 383Mbits/s
8 1024KB 247Mbits/s
4 2048KB 173Mbits/s
2 4096KB 87Mbits/s
1 8192KB 43Mbits/s
64 256KB 544Mbits/s #200 secs
Thus it appears that the host is not limited, it can push aside other TCP friendly traffic so it gets over 500Mbits/s achievable throughput.
Another interesting observation obtained from looking at iepmbw.cacr.caltech.edu data was that ping max Rtt showed a similar behavior ad path chirp throughput which forced us to think that ping can be the replacement of rather complex path chirp tool. The data from which the graph is generated is avialable here
After all this analysis we concluded that
1- Route change is not the cause
2- Network behavior/ Congestion is not the cause
3- Only possibility is that iepm-bw.bnl.gov was itself experiencing some bad performance which was causing some of the monitored nodes to show negative results.