Neo4j Is Whiteboard Friendly

Imagine you need to create a wine suggestion engine in which wines are categorized by different varieties, regions, wineries, vintages, and designations. Imagine that you also need to keep track of things like articles describing those wines written by various authors and to enable users to track their favorite wines.

If you were using a relational model, you might create a category table and a many-to-many relationship between a single winery’s wine and some combination of categories and other data. But this isn’t quite how humans mentally model data. In the following figure, compare this wine suggestion schema in relational UML:

to this wine suggestion data on a whiteboard:

There’s an old saying in the relational database world: on a long enough timeline, all fields become optional. Neo4j handles this implicitly by providing values and structure only where necessary. If a wine blend has no vintage, add a bottle year and point the vintages to the blend node instead. In graph databases such as Neo4j there is simply no schema to adjust.

Over the next three days you’ll learn how to interact with Neo4j through a web console, using a querying language called Cypher, then via a REST interface, and finally through search indexes. You’ll work with some simple graphs as well as some larger graphs with graph algorithms. Finally, on Day 3, you’ll take a peek at the enterprise tools that Neo4j provides for mission-critical applications, from full ACID-compliant transactions to high-availability clustering and incremental backups.

In this chapter, we’ll use the Neo4j 3.1.4 Enterprise Edition. Most of the actions you perform on Days 1 and 2 can actually use the GPL Community edition, but we’ll require some enterprise functionality for Day 3: Distributed High Availability. You can download a free trial version of the Enterprise Edition from the Neo4j website.