The ESCAPE code

This repository contains code for the ESCAPE counter scheme.

Scripts for CAIDA traces

Preprocessing

The scripts are intended to compute basic statistics for the ESCAPE counter scheme on CAIDA traces; they reproduce the figures shown in the CoNEXT 2016 submission "Exact Scalable Per-flow Packet Counting".

Before running the scripts, obtain CAIDA traces from http://www.caida.org/data/passive/passive_2016_dataset.xml (registration needed).

For each CAIDA trace trace.pcap.gz, run the following preprocessing:

gunzip trace.pcap.gz
tcpdump -qn -r trace.pcap > trace.txt
python3 conext2016_scripts/find_last_flow_times.py trace.txt

The find_last_flow_times.py script finds the last timestamp for each flow in the trace and dumps them to the file trace.lasttimes.txt; subsequent scripts will use these timestamps as the time when a flow closes (since there are no explicit flow-ending commands in the scripts).

Reproducing the figures

The directory conext2016_scripts contains Python scripts that recreate the graphs from the CoNEXT 2016 submission "Exact Scalable Per-flow Packet Counting". Each script expects a trace.txt file and a trace.lasttimes.txt file generated as above. All scripts output files in the form of figurename.trace.csv that contain X- and Y-axis values ready to input into pgfplots. The figures are grouped into a single script if they need the same processing (e.g., go over the trace once or re-run for all polling ratios).

Figure 5 ("Validation of the value-sharing phenomenon"), Figure 6a ("Counters and memory")

Run

python3 conext2016_scripts/conext_fig5_6a.py trace.txt

to produce Figs. 5a, 5b, 5c, and 6a for a given trace.

Figures 6b and 6c ("Counters and memory")

Run

python3 conext2016_scripts/conext_fig6b_6c.py trace.txt

to produce Figs. 6a and 6b for a given trace.

Figure 7a ("Polling configurations and ESCAPE processing complexity")

Run

python3 conext2016_scripts/conext_fig7a.py trace.txt

to produce Fig. 7a for a given trace.

Figure 7b ("Polling configurations and ESCAPE processing complexity")

Run

python3 conext2016_scripts/conext_fig7b.py trace.txt

to produce Fig. 7b for a given trace. Note that this will take longest to generate among all figures since it re-runs trace processing with different polling rates.

Figure 7c ("Polling configurations and ESCAPE processing complexity")

Run

python3 conext2016_scripts/conext_fig7c.py trace.txt

to produce Fig. 7c for a given trace. Note that this will take longer than the other figures (except 7b).