Changes between Version 3 and Version 4 of PlasticSlices/BaselineEvaluation/Baseline5Details


Ignore:
Timestamp:
06/15/11 14:45:43 (13 years ago)
Author:
Josh Smift
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • PlasticSlices/BaselineEvaluation/Baseline5Details

    v3 v4  
    44
    55We changed how we were capturing logs for this baseline, going back to screen's 'screenlog' function, to ensure that we had a more complete record of each experiment. The raw logs are at http://www.gpolab.bbn.com/plastic-slices/baseline-logs/baseline-5/.
     6
     7We made an error in the slices that use ICMP and UDP (plastic-101, plastic-102, plastic-104, and plastic-106): We ran commands to generate 1 Mb/sec of traffic, rather than ones to generate 10 Mb/sec.
    68
    79= plastic-101 =
     
    101103== Analysis ==
    102104
    103 ''(forthcoming)''
     105There were some outages at Stanford in here, which largely account for the high packet loss rates in the pairs that include Stanford.
     106
     107Wisconsin - Washington had an unexpectedly high loss rate.
     108
     109The other results seem consistent with what we'd expect.
    104110
    105111= plastic-102 =
     
    203209== Analysis ==
    204210
    205 ''(forthcoming)''
     211of-planet3.stanford.edu was effectively unusable during this baseline.
     212
     213Wisconsin - Washington had an unexpectedly high loss rate.
     214
     215The other results seem consistent with what we'd expect.
    206216
    207217= plastic-103 =
     
    375385== Analysis ==
    376386
    377 ''(forthcoming)''
     387There were some outages at Stanford in here, which accounts for the initial problems with Stanford - BBN and Stanford - Washington.
     388
     389of-planet3.stanford.edu was effectively unusable during this baseline.
     390
     391The other results seem consistent with what we'd expect.
    378392
    379393= plastic-104 =
     
    519533== Analysis ==
    520534
    521 ''(forthcoming)''
     535All results seem consistent with what we'd expect.
    522536
    523537= plastic-105 =
     
    618632== Analysis ==
    619633
    620 ''(forthcoming)''
     634GT - Wisconsin got disconnected at one point, which explains the much higher rate in the second run relative to the first.
     635
     636The other results seem consistent with what we'd expect.
    621637
    622638= plastic-106 =
     
    729745== Analysis ==
    730746
    731 ''(forthcoming)''
     747All results seem consistent with what we'd expect.
    732748
    733749= plastic-107 =
     
    828844== Analysis ==
    829845
    830 ''(forthcoming)''
     846All results seem consistent with what we'd expect.
    831847
    832848= plastic-108 =
     
    927943== Analysis ==
    928944
    929 ''(forthcoming)''
     945of-planet3.stanford.edu was effectively unusable during this baseline.
     946
     947The other results seem consistent with what we'd expect.
    930948
    931949= plastic-109 =
     
    10301048== Analysis ==
    10311049
    1032 ''(forthcoming)''
     1050All results seem consistent with what we'd expect.
    10331051
    10341052= plastic-110 =
     
    11331151== Analysis ==
    11341152
    1135 ''(forthcoming)''
     1153of-planet4.stanford.edu was actually usable in other experiments, but failed entirely in this one.
     1154
     1155The other results seem consistent with what we'd expect.