Changes between Version 2 and Version 3 of PlasticSlices/BaselineEvaluation/Baseline7Details


Ignore:
Timestamp:
07/11/11 17:25:46 (14 years ago)
Author:
Josh Smift
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • PlasticSlices/BaselineEvaluation/Baseline7Details

    v2 v3  
    33Here are the details of [wiki:PlasticSlices/BaselineEvaluation#Baseline7 Baseline 7], divided by slice.
    44
    5 In this baseline, we ran multiple rounds of experiments, one for each set of Emergency Stop tests. (One so far, more to come.)
     5In this baseline, we intended to run multiple rounds of experiments, one for each set of Emergency Stop tests, but we only ended up running one round. We left the "round" structure (on this page and in the log directory) in place for simplicity. We'll run additional rounds after GEC 11 in the continuation phase of this project.
    66
    77The raw logs of each experiment are at http://www.gpolab.bbn.com/plastic-slices/baseline-logs/baseline-7/, with a subdirectory for the logs from each round.
    88
     9The logs from this baseline are somewhat misleading, because we forgot to stop the experiments immediately after the Emergency Stop tests, so they show disruptions from the following day (e.g. from the NLR OpenFlow stack migration) that are entirely unrelated to the Emergency Stop tests themselves.
     10
    911= plastic-101 =
    1012
    11 [PlasticSlices/Experiments#GigaPing GigaPing], using count=100000, and this table of client/server pairs:
     13[PlasticSlices/Experiments#SteadyPing SteadyPing], using interval=.006, and this table of client/server pairs:
    1214
    1315|| '''client'''                   || '''server'''                   || '''server address''' ||
    … …  
    109111=== Analysis ===
    110112
    111 ''(forthcoming)''
    112 
    113 == Round 2 ==
    114 
    115 === Results ===
    116 
    117 Generated from the logs with
    118 
    119 {{{
    120 subnet=101
    121 for host in ganel.gpolab.bbn.com planetlab4.clemson.edu plnode1.cip.gatech.edu pl4.myplc.grnoc.iu.edu orbitplc1.orbit-lab.org of-planet1.stanford.edu pl01.cs.washington.edu wings-openflow-2.wail.wisc.edu ; do echo -e "$host:\n\n{{{" ; grep -a -A 2 statistics pgenigpolabbbncom_plastic$subnet\@$host.log ; echo -e "}}}\n" ; done
    122 }}}
    123 
    124 === Analysis ===
    125 
    126 ''(forthcoming)''
     113The high packet loss was caused by the NLR OpenFlow software stack migration, not the Emergency Stop test itsef.
    127114
    128115= plastic-102 =
    129116
    130 [PlasticSlices/Experiments#GigaPing GigaPing], using count=100000, and this table of client/server pairs:
     117[PlasticSlices/Experiments#SteadyPing SteadyPing], using interval=.006, and this table of client/server pairs:
    131118
    132119|| '''client'''                   || '''server'''                   || '''server address''' ||
    … …  
    229216=== Analysis ===
    230217
    231 ''(forthcoming)''
    232 
    233 == Round 2 ==
    234 
    235 === Results ===
    236 
    237 Generated from the logs with
    238 
    239 {{{
    240 subnet=102
    241 for host in sardis.gpolab.bbn.com planetlab5.clemson.edu plnode2.cip.gatech.edu pl5.myplc.grnoc.iu.edu orbitplc2.orbit-lab.org of-planet1.stanford.edu pl02.cs.washington.edu wings-openflow-3.wail.wisc.edu ; do echo -e "$host:\n\n{{{" ; grep -a -A 2 statistics pgenigpolabbbncom_plastic$subnet\@$host.log ; echo -e "}}}\n" ; done
    242 }}}
    243 
    244 === Analysis ===
    245 
    246 ''(forthcoming)''
     218The high packet loss was caused by the NLR OpenFlow software stack migration, not the Emergency Stop test itsef.
    247219
    248220= plastic-103 =
    249221
    250 [PlasticSlices/Experiments#GigaPerfTCP GigaPerf TCP], using port=5103, size=350, and this table of client/server pairs:
     222[PlasticSlices/Experiments#SteadyPerfTCP SteadyPerf TCP], using port=5103, time=518400, and this table of client/server pairs:
    251223
    252224|| '''client'''                   || '''server'''                   || '''server address''' ||
    … …  
    338310=== Analysis ===
    339311
    340 ''(forthcoming)''
    341 
    342 == Round 2 ==
    343 
    344 === Results ===
    345 
    346 Generated with
    347 
    348 {{{
    349 subnet=103
    350 for host in of-planet1.stanford.edu ganel.gpolab.bbn.com pl02.cs.washington.edu of-planet2.stanford.edu orbitplc2.orbit-lab.org ; do echo -e "$host:\n\n{{{" ; grep -a -A 5 -B 1 "Client connecting" pgenigpolabbbncom_plastic$subnet\@$host.log ; echo -e "}}}\n" ; done
    351 }}}
    352 
    353 and then edited slightly to remove artifacts (like control characters, my prompt, etc).
    354 
    355 === Analysis ===
    356 
    357 ''(forthcoming)''
     312The results are reported strangely: "14702427515809245184 bits  0.00 (null)s/sec"? The server-side logs make it clear that data was being transfered, though -- with some gaps, but by and large, as expected.
    358313
    359314= plastic-104 =
    360315
    361 [PlasticSlices/Experiments#GigaPerfUDP GigaPerf UDP], using port=5104, size=500, rate=100, and this table of client/server pairs:
     316[PlasticSlices/Experiments#SteadyPerfUDP SteadyPerf UDP], using port=5104, time=518400, rate=3, and this table of client/server pairs:
    362317
    363318|| '''client'''                   || '''server'''                   || '''server address''' ||
    … …  
    501456=== Analysis ===
    502457
    503 ''(forthcoming)''
    504 
    505 == Round 2 ==
    506 
    507 === Results ===
    508 
    509 Generated with
    510 
    511 {{{
    512 subnet=104
    513 for host in navis.gpolab.bbn.com wings-openflow-3.wail.wisc.edu orbitplc1.orbit-lab.org pl5.myplc.grnoc.iu.edu ; do echo -e "$host:\n\n{{{" ; grep -a -A 10 -B 1 "Client connecting" pgenigpolabbbncom_plastic$subnet\@$host.log ; echo -e "}}}\n" ; done
    514 }}}
    515 
    516 and then edited slightly to remove artifacts (like control characters, my prompt, etc).
    517 
    518 === Analysis ===
    519 
    520 ''(forthcoming)''
     458The results are generally consistent with what we'd expect.
    521459
    522460= plastic-105 =
    523461
    524 [PlasticSlices/Experiments#GigaPerfTCP GigaPerf TCP], using port=5105, size=250, and this table of client/server pairs:
     462[PlasticSlices/Experiments#SteadyPerfTCP SteadyPerf TCP], using port=5105, time=518400, and this table of client/server pairs:
    525463
    526464|| '''client'''                   || '''server'''                   || '''server address''' ||
    … …  
    623561=== Analysis ===
    624562
    625 ''(forthcoming)''
    626 
    627 == Round 2 ==
    628 
    629 === Results ===
    630 
    631 Generated with
    632 
    633 {{{
    634 subnet=105
    635 for host in wings-openflow-2.wail.wisc.edu planetlab4.clemson.edu bain.gpolab.bbn.com plnode1.cip.gatech.edu ; do echo -e "$host:\n\n{{{" ; grep -a -A 5 -B 1 "Client connecting" pgenigpolabbbncom_plastic$subnet\@$host.log ; echo -e "}}}\n" ; done
    636 }}}
    637 
    638 and then edited slightly to remove artifacts (like control characters, my prompt, etc).
    639 
    640 === Analysis ===
    641 
    642 ''(forthcoming)''
     563As in plastic-103, the results are reported strangely, but the server-side logs make it clear that data was being transfered.
    643564
    644565= plastic-106 =
    645566
    646 [PlasticSlices/Experiments#GigaPerfUDP GigaPerf UDP], using port=5106, size=500, rate=100, and this table of client/server pairs:
     567[PlasticSlices/Experiments#SteadyPerfUDP SteadyPerf UDP], using port=5106, time=518400, rate=3, and this table of client/server pairs:
    647568
    648569|| '''client'''                   || '''server'''                   || '''server address''' ||
    … …  
    751672}}}
    752673
    753 == Round 2 ==
    754 
    755 === Results ===
    756 
    757 Generated with
    758 
    759 {{{
    760 subnet=106
    761 for host in planetlab5.clemson.edu wings-openflow-3.wail.wisc.edu plnode2.cip.gatech.edu gardil.gpolab.bbn.com ; do echo -e "$host:\n\n{{{" ; grep -a -A 10 -B 1 "Client connecting" pgenigpolabbbncom_plastic$subnet\@$host.log ; echo -e "}}}\n" ; done
    762 }}}
    763 
    764 and then edited slightly to remove artifacts (like control characters, my prompt, etc).
    765 
    766674=== Analysis ===
    767675
    768 ''(forthcoming)''
     676The results are generally consistent with what we'd expect.
    769677
    770678= plastic-107 =
    771679
    772 [PlasticSlices/Experiments#GigaWeb GigaWeb], using count=30, port=4107, file=substrate.doc, md5sum=d4fcf71833327fbfef98be09deef8bfb, and this table of client/server pairs:
     680[PlasticSlices/Experiments#SteadyWeb SteadyWeb], using port=4107, file=substrate.doc, md5sum=d4fcf71833327fbfef98be09deef8bfb, and this table of client/server pairs:
    773681
    774682|| '''client'''                   || '''server'''                   || '''server address''' ||
    … …  
    867775=== Analysis ===
    868776
    869 ''(forthcoming)''
    870 
    871 == Round 2 ==
    872 
    873 === Results ===
    874 
    875 Since we transfered the file to each client hundreds (if not thousands) of times, we didn't save all the copies, but we did generate an MD5 checksum after each download, and log that, so we ran some grep commands on the logs to count the number of total checksums printed in each log, and the number where the checksum had the expected value.
    876 
    877 Generated with
    878 
    879 {{{
    880 subnet=107
    881 for host in planetlab5.clemson.edu pl5.myplc.grnoc.iu.edu plnode1.cip.gatech.edu pl01.cs.washington.edu ; do echo -e -n "$host:\n\n{{{\ntotal md5sum lines: " ; grep -a "md5sum:" pgenigpolabbbncom_plastic$subnet\@$host.log | wc -l ; echo -e -n "matching md5sum lines: " ; grep -a "md5sum: d4fcf71833327fbfef98be09deef8bfb" pgenigpolabbbncom_plastic$subnet\@$host.log | wc -l ; echo -e "}}}\n" ; done
    882 }}}
    883 
    884 === Analysis ===
    885 
    886 ''(forthcoming)''
     777The results are generally consistent with what we'd expect.
    887778
    888779= plastic-108 =
    889780
    890 [PlasticSlices/Experiments#GigaWeb GigaWeb], using count=40, port=4108, file=substrate.doc, md5sum=d4fcf71833327fbfef98be09deef8bfb, and this table of client/server pairs:
     781[PlasticSlices/Experiments#SteadyWeb SteadyWeb], using port=4108, file=substrate.doc, md5sum=d4fcf71833327fbfef98be09deef8bfb, and this table of client/server pairs:
    891782
    892783|| '''client'''                   || '''server'''                   || '''server address''' ||
    … …  
    985876=== Analysis ===
    986877
    987 ''(forthcoming)''
    988 
    989 == Round 2 ==
    990 
    991 === Results ===
    992 
    993 Since we transfered the file to each client hundreds (if not thousands) of times, we didn't save all the copies, but we did generate an MD5 checksum after each download, and log that, so we ran some grep commands on the logs to count the number of total checksums printed in each log, and the number where the checksum had the expected value.
    994 
    995 Generated with
    996 
    997 {{{
    998 subnet=108
    999 for host in wings-openflow-3.wail.wisc.edu of-planet1.stanford.edu orbitplc2.orbit-lab.org pl4.myplc.grnoc.iu.edu ; do echo -e -n "$host:\n\n{{{\ntotal md5sum lines: " ; grep -a "md5sum:" pgenigpolabbbncom_plastic$subnet\@$host.log | wc -l ; echo -e -n "matching md5sum lines: " ; grep -a "md5sum: d4fcf71833327fbfef98be09deef8bfb" pgenigpolabbbncom_plastic$subnet\@$host.log | wc -l ; echo -e "}}}\n" ; done
    1000 }}}
    1001 
    1002 === Analysis ===
    1003 
    1004 ''(forthcoming)''
     878The results are generally consistent with what we'd expect.
    1005879
    1006880= plastic-109 =
    1007881
    1008 [PlasticSlices/Experiments#GigaNetcat GigaNetcat], using count=25, port=6109, file=substrate.doc, and this table of client/server pairs:
     882[PlasticSlices/Experiments#SteadyNetcat SteadyNetcat], using port=6109, file=substrate.doc, md5sum=d4fcf71833327fbfef98be09deef8bfb, and this table of client/server pairs:
    1009883
    1010884|| '''client'''                   || '''server'''                   || '''server address''' ||
    … …  
    1107981=== Analysis ===
    1108982
    1109 ''(forthcoming)''
    1110 
    1111 == Round 2 ==
    1112 
    1113 === Results ===
    1114 
    1115 Since we transfered the file to each client hundreds (if not thousands) of times, we didn't save all the copies, but we did generate an MD5 checksum after each download, and log that, so we ran some grep commands on the logs to count the number of total checksums printed in each log, and the number where the checksum had the expected value.
    1116 
    1117 Generated with
    1118 
    1119 {{{
    1120 subnet=109
    1121 for host in navis.gpolab.bbn.com pl4.myplc.grnoc.iu.edu pl01.cs.washington.edu planetlab4.clemson.edu of-planet2.stanford.edu wings-openflow-2.wail.wisc.edu ; do echo -e -n "$host:\n\n{{{\ntotal md5sum lines: " ; grep -a "substrate.doc" pgenigpolabbbncom_plastic$subnet\@$host.log | wc -l ; echo -e -n "matching md5sum lines: " ; grep -a "d4fcf71833327fbfef98be09deef8bfb  substrate.doc" pgenigpolabbbncom_plastic$subnet\@$host.log | wc -l ; echo -e "}}}\n" ; done
    1122 }}}
    1123 
    1124 === Analysis ===
    1125 
    1126 ''(forthcoming)''
     983The results are generally consistent with what we'd expect.
    1127984
    1128985= plastic-110 =
    1129986
    1130 [PlasticSlices/Experiments#GigaNetcat GigaNetcat], using count=25, port=6110, file=substrate.doc, and this table of client/server pairs:
     987[PlasticSlices/Experiments#SteadyNetcat SteadyNetcat], using port=6110, file=substrate.doc, md5sum=d4fcf71833327fbfef98be09deef8bfb, and this table of client/server pairs:
    1131988
    1132989|| '''client'''                   || '''server'''                   || '''server address''' ||
    … …  
    12291086=== Analysis ===
    12301087
    1231 ''(forthcoming)''
    1232 
    1233 == Round 2 ==
    1234 
    1235 === Results ===
    1236 
    1237 Since we transfered the file to each client hundreds (if not thousands) of times, we didn't save all the copies, but we did generate an MD5 checksum after each download, and log that, so we ran some grep commands on the logs to count the number of total checksums printed in each log, and the number where the checksum had the expected value.
    1238 
    1239 Generated with
    1240 
    1241 {{{
    1242 subnet=110
    1243 for host in gardil.gpolab.bbn.com pl02.cs.washington.edu of-planet2.stanford.edu orbitplc1.orbit-lab.org pl5.myplc.grnoc.iu.edu plnode2.cip.gatech.edu ; do echo -e -n "$host:\n\n{{{\ntotal md5sum lines: " ; grep -a "substrate.doc" pgenigpolabbbncom_plastic$subnet\@$host.log | wc -l ; echo -e -n "matching md5sum lines: " ; grep -a "d4fcf71833327fbfef98be09deef8bfb  substrate.doc" pgenigpolabbbncom_plastic$subnet\@$host.log | wc -l ; echo -e "}}}\n" ; done
    1244 }}}
    1245 
    1246 === Analysis ===
    1247 
    1248 ''(forthcoming)''
     1088The results are generally consistent with what we'd expect.