Localised bottlenecks for server transport layer protocol analysis.
- Ayush Mishra, Xiangpeng Sun, Atishya Jain, Sameer Pande, Raj Joshi, and Ben Leong. "The Great Internet TCP Congestion Control Census". Proceedings of the ACM on Measurement and Analysis of Computing Systems (SIGMETRICS 2019). Volume 3. Issue 3. December 2019.
Scripts/quick_dependancies.sh
- Launch a delay shell
$ mm-delay 50 - The scripts for running the necessary tests are in Scripts/
mm-delay 50 $ cd Scripts/ - Launching the test
mm-delay 50 $ ./launch.sh <target_URL> - Ending all processes (not necessary)
mm-delay 50 $ ./clean.sh
- Launch a delay shell
$ mm-delay 50 - The scripts for running the necessary tests are in Scripts/
mm-delay 50 $ cd Scripts/ - Launching the test
mm-delay 50 $ ./multi-launch.sh <target_URL> <number of trials per RTT>
each trial's data will be stored in Gordon/Data/windows<trail_num>.csv
cleaned data is stored in Gordon/Data/windows.csv - Ending all processes (not necessary)
mm-delay 50 $ ./clean.sh
The final data at the end of either of the above tests are stored in windows.csv. The individual trials for 'multiple-trails-per-RTT' test type are stored as windows1.csv, windows2.csv, ... windowsn.csv for 'n' trials.
Plot the results using gnuplot plot.plt --persist for multiple-trials tests.
For single trial tests, you can simply open a gnuplpot shell and plot as plot 'windows.csv' u 3:2 w lines