Artifact Evaluation for Final Year Project (FYP)

Overview

This repository is for artifact evaluation (i.e., 软硬件验收 in Chinese) of final year project (FYP) in Xidian University.

Section 2 contains necessary steps to set up the testbed to reproduce the result. Section 3 gives instruction to generate the trace used in the evaluation. Section 4 provides detailed instruction to reproduce the result in bachelor thesis.

Setup environment

Testbed Overview:

Install Dependencies:

pip install scapy
sudo apt-get install libpcap-dev linux-tools-5.4.0-214-generic

System configuration:

# for perf
cat /proc/sys/kernel/perf_event_paranoid
echo "-1" | sudo  tee /proc/sys/kernel/perf_event_paranoid
# enable turbo boost
cat /sys/devices/system/cpu/intel_pstate/no_turbo
echo "0" | sudo tee /sys/devices/system/cpu/intel_pstate/no_turbo

# Two NUMA node here
# Hugepage setup
echo 2048 | sudo tee /sys/devices/system/node/node0/hugepages/hugepages-2048kB/nr_hugepages
echo 2048 | sudo tee /sys/devices/system/node/node1/hugepages/hugepages-2048kB/nr_hugepages
mkdir /mnt/huge
mount -t hugetlbfs pagesize=1GB /mnt/huge

All the programs are developed and tested with DPDK 21.05.

DPDK setup (Note that you must replaced the 0000:82:00.1 and 0000:04:00.1 with your configuration.):

# add the NIC used in experiment
sudo dpdk-devbind.py -b vfio-pci 0000:82:00.1
sudo dpdk-devbind.py -b vfio-pci 0000:04:00.1
pkg-config --modversion libdpdk # show dpdk version

Compile the program used in motivation section:

make -C ./motivation

Compile the program used in evaluation section (including aggregator implementation):

make -C ./aggregator

Generate packets trace and configuration file for firewall

Generate packets trace for motivation and evaluation:

# machine time: ~40 min (can be parallelized to speed up)
./scripts/run_motivation.py --prepare
# machine time: ~6 min
./scripts/run_evaluation.py --prepare

Reproducing the results

Results for motivation

# machine time: ~ 35 minutes
./scripts/run_motivation.py --run

All generated figures are under result/fig.

Fig in Thesis	File Name
Fig 3-2	flow_num10000per_packet.png
Fig 3-3	flow_num10000miss_rate.png
Fig 3-4	flow_num10000per_packet.png
Fig 3-5	flow_num1000miss_rate.png
Fig 3-6	flow_num32per_packet.png

Results for evaluation

# machine time: ~17 minutes
sudo ./scripts/run_evaluation.py --run --buffer-time

All generated figures are under result/fig.

Fig in Thesis	File Name
Fig 4-3	throughput.png
Fig 4-4	latency.png
Fig 4-5	cycle.png
Fig 4-6	buffer_time_throughput.png
Fig 4-7	buffer_time_latency.png
Fig 4-8	buffer_time_cycle.png

Clean up all the results and traces

# remove results
rm -rf ./result
# remove traces 
rm -rf *.pcap
# To restore the original configuration
sudo dpdk-devbind.py -u 0000:04:00.1
sudo dpdk-devbind.py -u 0000:82:00.1
sudo dpdk-devbind.py -b i40e 0000:04:00.1
sudo dpdk-devbind.py -b i40e 0000:82:00.1