runme

#!/bin/bash
#Thomas Steinke (tsteinke@seas.harvard.edu) November 2010
#This README/runme file shows the overall structure of our implementation
#and experiments. Running this script will download the packet trace,
#preprocess it, compile our code, run experiments, and, finally, produce
#the graphs used in our paper. Running this script will take approximately
#three days. It is not meant to be entirely practical, but it makes it clear
#how our code works.
#
#Firstly,
#hhh = our algorithm
#uhhh = our algorithm with unitary updates
#omp = the parallel version of our algorithm
#ancestry = partial ancestry algorithm
#full = full ancestry algorithm
#Program names are appended by the dimension:
# *1 and *2 are byte-wise (1- and 2-d resp.) and 1_33 are bit-wise (1-d)
#
#There is a makefile that compiles all our programs.
make all
#the url of the trace we used. This is a 6.8GB download and is the largest trace we could find.
FILEURL=https://data.caida.org/datasets/oc48/oc48-original/20030115/1hour/20030115-100000-1-anon.pcap.gz
#need username and password from caida.org
#ATTENTION: THIS SCRIPT WILL FAIL WITHOUT A USERNAME AND PASSWORD. PLEASE VISIT www.caida.org.
USERPASS=user@example.com:Password
#download
curl -u $USERPASS $FILEURL > data.pcap.gz
#data needs to be made human-readable (takes time)
gunzip data.pcap.gz
tcpdump -q -n -t -r data.pcap | process > data.txt 2> /dev/null
rm data.pcap #saves 15GB
#now experiments can be done
#For set values use the following
#NUM_POINTS=1000000
#NUM_COUNTERS=1000
#THRESHOLD=1000
#./runtests $NUM_POINTS $NUM_COUNTERS $THRESHOLD data.txt
#data will appear in data.txt.summary
#alltests does a whole bunch of predetermined values
time ./alltests data.txt
echo "Results:"
cat data.txt.summary
#The following script will produce a lot of plots
./makeplot.py