Changes between Initial Version and Version 1 of PlasticSlices/Continuation/Round5

Timestamp:

03/07/12 10:51:07 (12 years ago)

Author:

Josh Smift

Comment:

--

PlasticSlices/Continuation/Round5

@@ +1 @@
+[[PageOutline]]
+
+= Continuation Round 5 =
+
+In this round, I ran plastic-105 and plastic-106 for about an hour, with only GT and Wisconsin, to try to track down some issues with performance in Round 3 that seemed related to GT. In particular, we wanted to find out whether there was any difference when GT was sending data vs receiving data.
+
+The raw logs are at http://www.gpolab.bbn.com/plastic-slices/continuation/round-5/. 
+
+= plastic-105 =
+
+[ggw:PlasticSlices/Experiments#SteadyPerfTCP SteadyPerf TCP], using port=5105, time=518400, and this table of client/server pairs:
+
+|| '''client'''                   || '''server'''                   || '''server address''' ||
+|| wings-openflow-2.wail.wisc.edu || plnode1.cip.gatech.edu         || server=10.42.105.100 ||
+|| wings-openflow-3.wail.wisc.edu || plnode2.cip.gatech.edu         || server=10.42.105.101 ||
+
+== One-time prep commands run on each client and server ==
+
+{{{
+sudo yum -y install iperf
+}}}
+
+== Commands run on each server ==
+
+{{{
+server=<ipaddr>
+nice -n 19 iperf -B $server -p 5105 -s -i 1
+}}}
+
+== Commands run on each client ==
+
+{{{
+server=<ipaddr>
+nice -n 19 iperf -c $server -p 5105 -t 518400
+}}}
+
+== Results ==
+
+Generated with
+
+{{{
+subnet=105
+for host in $(awk 'NR%2==1' ~/plastic-slices/logins/logins-plastic-$subnet.txt | sed -r -e 's/.+@//') ; do echo -e "$host:\n\n{{{" ; grep -A 5 -B 1 "Client connecting" pgenigpolabbbncom_plastic$subnet\@$host.log ; echo -e "}}}\n" ; done
+}}}
+
+and then edited slightly to remove artifacts (like control characters, my prompt, etc).
+
+wings-openflow-2.wail.wisc.edu:
+
+{{{
+------------------------------------------------------------
+Client connecting to 10.42.105.100, TCP port 5105
+TCP window size: 16.0 KByte (default)
+------------------------------------------------------------
+[  3] local 10.42.105.95 port 55411 connected with 10.42.105.100 port 5105
+[ ID] Interval       Transfer     Bandwidth
+[  3]  0.0-4904.3 sec  1.19 GBytes  2.09 Mbits/sec
+}}}
+
+wings-openflow-3.wail.wisc.edu:
+
+{{{
+------------------------------------------------------------
+Client connecting to 10.42.105.101, TCP port 5105
+TCP window size: 16.0 KByte (default)
+------------------------------------------------------------
+[  3] local 10.42.105.96 port 52296 connected with 10.42.105.101 port 5105
+[ ID] Interval       Transfer     Bandwidth
+[  3]  0.0-4903.6 sec  1.18 GBytes  2.06 Mbits/sec
+}}}
+
+== Analysis ==
+
+Not great numbers, but a factor of seven or eight times better than the other direction. This smells like a duplex problem to me -- remember that traffic is sent from the client to the server, in iperf.
+
+= plastic-106 =
+
+[ggw:PlasticSlices/Experiments#SteadyPerfUDP SteadyPerf UDP], using port=5106, time=518400, rate=30, and this table of client/server pairs:
+
+|| '''client'''                   || '''server'''                   || '''server address''' ||
+|| wings-openflow-2.wail.wisc.edu || plnode2.cip.gatech.edu         || server=10.42.105.101 ||
+|| wings-openflow-3.wail.wisc.edu || plnode1.cip.gatech.edu         || server=10.42.105.100 ||
+
+== One-time prep commands run on each client and server ==
+
+{{{
+sudo yum -y install iperf
+}}}
+
+== Commands run on each server ==
+
+{{{
+server=<ipaddr>
+nice -n 19 iperf -u -B $server -p 5106 -s -i 1
+}}}
+
+== Commands run on each client ==
+
+{{{
+server=<ipaddr>
+nice -n 19 iperf -u -c $server -p 5106 -t 518400 -b 30M
+}}}
+
+== Results ==
+
+Generated with
+
+{{{
+subnet=106
+for host in $(awk 'NR%2==1' ~/plastic-slices/logins/logins-plastic-$subnet.txt | sed -r -e 's/.+@//') ; do echo -e "$host:\n\n{{{" ; grep -A 10 -B 1 "Client connecting" pgenigpolabbbncom_plastic$subnet\@$host.log ; echo -e "}}}\n" ; done
+}}}
+
+and then edited slightly to remove artifacts (like control characters, my prompt, etc).
+
+wings-openflow-2.wail.wisc.edu:
+
+{{{
+------------------------------------------------------------
+Client connecting to 10.42.105.101, UDP port 5106
+Sending 1470 byte datagrams
+UDP buffer size:  110 KByte (default)
+------------------------------------------------------------
+[  3] local 10.42.105.95 port 50439 connected with 10.42.105.101 port 5106
+[ ID] Interval       Transfer     Bandwidth
+[  3]  0.0-4900.4 sec  17.1 GBytes  30.0 Mbits/sec
+[  3] Sent 12501122 datagrams
+[  3] Server Report:
+[  3]  0.0-4899.9 sec  17.1 GBytes  30.0 Mbits/sec   0.613 ms 6997/12501121 (0.056%)
+[  3]  0.0-4899.9 sec  161 datagrams received out-of-order
+}}}
+
+wings-openflow-3.wail.wisc.edu:
+
+{{{
+------------------------------------------------------------
+Client connecting to 10.42.105.100, UDP port 5106
+Sending 1470 byte datagrams
+UDP buffer size:  110 KByte (default)
+------------------------------------------------------------
+[  3] local 10.42.105.96 port 48014 connected with 10.42.105.100 port 5106
+[ ID] Interval       Transfer     Bandwidth
+[  3]  0.0-4903.0 sec  17.1 GBytes  30.0 Mbits/sec
+[  3] Sent 12507660 datagrams
+[  3] Server Report:
+[  3]  0.0-4902.8 sec  17.1 GBytes  30.0 Mbits/sec   0.852 ms 9669/12507659 (0.077%)
+[  3]  0.0-4902.8 sec  344 datagrams received out-of-order
+}}}
+
+== Analysis ==
+
+Not perfect, but three nines isn't bad. A significant difference over the other direction, anyway.