shixuansun / GraphStorageExp

Dynamic Graph Storage Survey

This repository provides source code and script in our experiments for reproducing the experiment in paper, Dynamic Graph Storage: An Experimental Survey.

We conduct an extensive survey and evaluation of existing graph storage systems, focusing on how the storage structures affect different aspects of performance, e.g., update throughput, lookup efficiency, algorithm execution time, and memory consumption.

All the experiment systems are open-source, and you can fetch them by the following links :

Aspen: https://github.com/ldhulipala/aspen

LLAMA: https://github.com/goatdb/llama

Teseo: https://github.com/cwida/teseo

Terrace: https://github.com/PASSIONLab/terrace

Stinger: https://github.com/the-data-lab/GraphOne

GraphOne: https://github.com/the-data-lab/GraphOne

LiveGraph: https://github.com/thu-pacman/LiveGraph

RisGraph: https://github.com/thu-pacman/RisGraph

PCSR: https://github.com/wheatman/Packed-Compressed-Sparse-Row/

Build

Environment

• CMake & C++ 17
• G++ 9.4.0 & gcc 7.5.0
• OpenMP
• Clang 10

Preconfig

git clone https://github.com/xiangyuzhi/GraphStorageExp.git --recursive
cd GraphStorageExp & mkdir build 
cd build
cmake ..
make

Dataset

The dataset can be download form: http://konect.cc/networks/ use:

cd data/LDBCgraph
wget http://konect.cc/networks/{exp_data} 

Before run experiment, please transform data from LDBC format to CSR format by use:

cd data/LDBCgraph
wget http://konect.cc/networks/{exp_data} 
cd ../../build
./transfer  ../data/LDBCgraph/{exp_data}.txt  ../data/ADJgraph/{exp_data}.adj

Run

The experiment for all system can be run following the script:

cd exp
./exp.sh

Plot

The figure plot in paper can be reproduced by running the python script in exp/plot/exp_data_proc.ipynb

Authors

• Xiangyu Zhi [e-mail]
• Xiao Yan [e-mail]
• Keming Li [e-mail]
• Bo Tang [e-mail]
• Yanchao Zhu [e-mail]
• Minqi Zhou [e-mail]