Dissect TCP traces in 20210202T170020.638359Z-pcap-mlab2-lga05-ndt.tgz
(See extraction directions below)
205 total traces
UUID between ndt-nnwk2_1611335823_00000000000C2D9F and ndt-nnwk2_1611335823_00000000000C2EFC
Some odd and example traces
ndt-nnwk2_1611335823_00000000000C2DF3 repeated timeouts for Verizon FIOS
ndt-nnwk2_1611335823_00000000000C2DDE/b2a_tsg.xpl Heavily thinned ACKS
ndt-nnwk2_1611335823_00000000000C2E75/b2a_tsg.xpl lossless variable rate, compressed ACKs (WiFI?)
ndt-nnwk2_1611335823_00000000000C2E85/b2a_tsg.xpl dirty WiFi
ndt-nnwk2_1611335823_00000000000C2DD8/b2a_tsg.xpl Fastest - has receiver pauses 480 Mb/s, not well paced
ndt-nnwk2_1611335823_00000000000C2E81/b2a_tsg.xpl 94+Mb/s, 10 Ms batched acks
ndt-nnwk2_1611335823_00000000000C2DAA/b2a_tsg.xpl repeated timeouts for Verizon FIOS
ndt-nnwk2_1611335823_00000000000C2DD0 FIOS with ECN
ndt-nnwk2_1611335823_00000000000C2DE8 FIOS with TSO???
ndt-nnwk2_1611335823_00000000000C2E15 FIOS receiver window limited
ndt-nnwk2_1611335823_00000000000C2E31 700+ Mb/s high loss short RTT
ndt-nnwk2_1611335823_00000000000C2E42 FIOS part ACK starved/thinned
ndt-nnwk2_1611335823_00000000000C2E74 FIOS ECN
ndt-nnwk2_1611335823_00000000000C2E91 really goofy looking win
ndt-nnwk2_1611335823_00000000000C2EF9 FIOS smooth 280k batching 575 Mb/s CUBIC ndt5
Fetch and extract the files
(reconstructed so I may have missed something.)
# fetch and extract
gsutil cp gs://archive-mlab-oti/ndt/pcap/2021/02/02/20210202T170020.638359Z-pcap-mlab2-lga05-ndt.tgz .
tar xvf 20210202T170020.638359Z-pcap-mlab2-lga05-ndt.tgz
cd 2021/02/02/
# put each trace in a separate dir
for f in *.gz ; do bn=`basename $f .pcap.gz`; mkdir $bn; mv $f $bn; done
# Run Sean Osterman's tcptrace on each (but is seems to have fallen off the web)
for d in ndt*; do
pushd $d
tcptrace -l -G -zxy *.pcap.gz > summary.txt
popd
done
# display one interesting trace (from the list above)
# Tim Shepeard's xplot
xplot ndt-nnwk2_1611335823_00000000000C2E81/b2a_tsg.xpl
# display all fios traces
for d in * ; do if fgrep -q fios $d/summary.txt ; then echo $d ; xplot $d/b2a_tsg.xpl ; fi ; done
Dissect TCP traces in 20210202T170020.638359Z-pcap-mlab2-lga05-ndt.tgz (See extraction directions below) 205 total traces UUID between ndt-nnwk2_1611335823_00000000000C2D9F and ndt-nnwk2_1611335823_00000000000C2EFC
Some odd and example traces ndt-nnwk2_1611335823_00000000000C2DF3 repeated timeouts for Verizon FIOS ndt-nnwk2_1611335823_00000000000C2DDE/b2a_tsg.xpl Heavily thinned ACKS ndt-nnwk2_1611335823_00000000000C2E75/b2a_tsg.xpl lossless variable rate, compressed ACKs (WiFI?) ndt-nnwk2_1611335823_00000000000C2E85/b2a_tsg.xpl dirty WiFi ndt-nnwk2_1611335823_00000000000C2DD8/b2a_tsg.xpl Fastest - has receiver pauses 480 Mb/s, not well paced ndt-nnwk2_1611335823_00000000000C2E81/b2a_tsg.xpl 94+Mb/s, 10 Ms batched acks ndt-nnwk2_1611335823_00000000000C2DAA/b2a_tsg.xpl repeated timeouts for Verizon FIOS ndt-nnwk2_1611335823_00000000000C2DD0 FIOS with ECN ndt-nnwk2_1611335823_00000000000C2DE8 FIOS with TSO??? ndt-nnwk2_1611335823_00000000000C2E15 FIOS receiver window limited ndt-nnwk2_1611335823_00000000000C2E31 700+ Mb/s high loss short RTT ndt-nnwk2_1611335823_00000000000C2E42 FIOS part ACK starved/thinned ndt-nnwk2_1611335823_00000000000C2E74 FIOS ECN ndt-nnwk2_1611335823_00000000000C2E91 really goofy looking win ndt-nnwk2_1611335823_00000000000C2EF9 FIOS smooth 280k batching 575 Mb/s CUBIC ndt5
Fetch and extract the files
(reconstructed so I may have missed something.)