Figure 8

Snort-filter (VM)

hipnet21/figures/snort-filter-nohz-tlbshootdown/vm/pcap/latencies-pre-rate10000-snort-filter.pcap.worst-filtered.pdf.svg

Steps to reproduce measurements

The setup and experiment execution for the worst-case measurements of Snort-fwd (HW) evaluation is part of the description for Figure 5. The high latency spike caused by TLB shootdowns happens on the first or second measurement after the machine has booted.

Steps to reproduce figures

Raw data

Pcaps (Snort-filter (VM))

Evaluation execution

Setup evaluator

# install dependencies
apt install postgresql postgresql-client parallel python3-pip texlive-full zstd
python3 -m pip install pypacker

Generate figures

git clone https://github.com/gallenmu/hipnet21 /root/hipnet21

#decompress pcaps
cd /root/hipnet21/measurements
for i in $(find . -iname *.zst); do unzstd --rm $i; done;

#create hw figure
export MPATH=snort-filter-nohz-tlbshootdown
cd /root/hipnet21/figures/$MPATH/vm/pcap
su postgres -c "createuser -s root"
dropdb --if-exists root
createdb root
export PGDATABASE=root
for i in /root/hipnet21/measurements/$MPATH/vm/pcap/*-pre-rate10000-*.pcap; do /root/hipnet21/scripts/pcap-import.sh $i; /root/hipnet21/scripts/pcap-analysis.sh $i; done;
python3 /root/hipnet21/scripts/pcap-tex-create.py .
for i in *.tex; do lualatex $i; done

Output

The precompiled output files can be found in the respective subfolders of the repository. The filename is latencies-pre-rate10000-snort-filter.pcap.worst-filtered.pdf in the respective subfolder.