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/
- CMake & C++ 17
- G++ 9.4.0 & gcc 7.5.0
- OpenMP
- Clang 10
git clone https://github.com/xiangyuzhi/GraphStorageExp.git --recursive
cd GraphStorageExp & mkdir build
cd build
cmake ..
make
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
The experiment for all system can be run following the script:
cd exp
./exp.sh
The figure plot in paper can be reproduced by running the python script in exp/plot/exp_data_proc.ipynb
- Xiangyu Zhi zhixy2021@mail.sustech.edu.cn
- Xiao Yan yanx@sustech.edu.cn
- Keming Li likm2020@mail.sustech.edu.cn
- Bo Tang tangb3@sustech.edu.cn
- Yanchao Zhu zhuyanchao2@huawei.com
- Minqi Zhou zhouminqi@huawei.com