mikebusuttil / graphdb-playground

Northwind on RDBMS (MariaDB) vs Graph Database (Neo4j)

This project was created to compare working with the same data using 2 different storage paradigms: graph database versus relational database.

Getting started

• Browse the graph-migrated Northwind dataset by visiting http://northwind.techiteasy.ca:7474 with username neo4j and password password
• Query the Northwind dataset in its original form with a MySQL client (ie. MySQL Workbench) by connecting to the following:
  • Hostname: northwind.techiteasy.ca
  • port: 3310
  • Username: mariadb
  • Password: password

DIY

To self-host the databases, refer to the following:

pre-requisites

The following assumes you have docker engine & docker-compose installed. Also, the MySQL root password is taken from your system's environment variable "NORTHWIND_MYSQL" which should be set in order for steps #1 & #2 below to work.

building the environment

1. start the database engines by checking out this repo and running the following from the repo's root:

docker-compose -f ./environment/dbz.yml up -d

• note: when running docker-compose as root you may lose permission to the data directory. To fix this run:

sudo chmod -R 777 ./environment/data

2. populate the relational database with the following:

docker exec -i nice_rdbms mysql -u root --password=$NORTHWIND_MYSQL < ./environment/data/northwind.sql

3. populate the graph database with the following:

docker exec -i nice_graph /var/lib/neo4j/bin/cypher-shell -u neo4j -p password < ./environment/data/import_csv.cypher

tearing down the environment

docker-compose -f ./environment/dbz.yml down
sudo rm -rf ./environment/sql
sudo rm -rf ./environment/cql

2do

• expand on the exercises & answers sections to get a better sense of how working with the 2 paradigms differs:
  • devise a query that spans all tables
• find and import the data for customer demographics.
• import the data for region, territory, employee territory, customer demographics, and state into Neo4j.
• format the numbers nicely (ie. $3,002.67) in neo4j queries - this might involve the installation of apoc.
• generate a very large data set to see how the performance of each paradigm scales.
• implement CI infrastructure that checks SQL answers against CQL answers and flags mis-matches in PR's.

References

• Many of the exercises were adapted from SQL Practice Problems: 57 beginning, intermediate, and advanced challenges for you to solve using a “learn-by-doing” approach by Sylvia Moestl Vasilik