graph-relational-benchmark

A fork to use docker for the deployment, there will be three containers :

• the benchmark web application
• the neo4j database
• the sql server database It will be launch through a docker-compose

This project enables you to benchmark graph and relational databases to compare their performance.

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.

Prerequisites

What things you need to install the software and how to install them:

To be able to connect to the Microsoft SQL database that will be created using the creation scripts below, there is a need to set up a Microsoft SQL Server on your local computer.

Neo4j Desktop Client also needs to be installed to be able to query the created Neo4j database, it can be downloaded here.

Project setup

Customize configuration

See Configuration Reference.

Installing

A step by step series of examples that tell you how to get a development env running

Install required dependencies for both python and node

pip install -r requirements.txt
npm install

Add environment variables into a .env file

# .env
SQL_SERVER=XXXXXX (e.g. 8ZC5G31)

Initiate the databases and populate them with data

To create the Microsoft SQL Server database, run the following

sqlcmd -i "backend/api/db/out/sql/create_db.sql"

Neo4j databases are easiest created in the Neo4j desktop client by clicking 'Projects' -> 'New' -> 'Add Graph' -> 'Start'.

The web app will automatically connect to the Neo4j database that is currently running.

! Important: Remove (comment out) the config option: dbms.directories.import=import from Neo4j -> Database -> Settings before population the graph database with fake data.

To populate the databases with fake data, run the following

npm run setup

This script runs reset_sql.sh -u && reset_cypher.sh in the backend/api/db folder.

Note: Only the first data population command needs the argument -u, as it creates the data which will be used to populate both databases. Nothing bad will happen if -u is provided to both scripts, the process will just take twice as long.

Starting the web app and the django server

To start the development server, run

npm start

which will start the web app at http://localhost:8080 as well as starting the backend server.

In the web app, pressing a tab and the 'refresh'-button will run the query shown at the bottom of the page a given amount of times.

The amount of query runs can be specified through the tab 'Configuration' -> Specify Query Amount -> 'Save'.

Note: Some queries, e.g. the queries in the 'documents' and 'histories' tab take a very long time and can therefore only be run a few times without the request timing out.

Query Complexity Analysis

To analyse the complexity of the queries that are used for the benchmark, run the following command:

./complexity_analyzer.sh cypher_argument sql_argument

Where cypher_argument and sql_argument are the queries that are to be compared, e.g.

./complexity_analyzer.sh match select

will compare the complexity of the cypher match queries that correspond to the select queries in SQL.

Built With

• Vue.js - The front-end web framework used
• Django - The back-end web framework used

Authors

• Jan Zubac - JanZubac
• Victor Winberg - VictorWinberg

See also the list of contributors who participated in this project.

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Acknowledgments