Graph databases: OrientDB to the rescue

OrientDB is a graph database written in Java, mainly developed by Luca Garulli, AssetData’s CTO.

Why a graph database

Sometimes the relation model isn’t the way you want both to collect and navigate your data: Luca Bonmasser, at the NoSQL day, highlighted it in his talk about anti-patterns in the relational ecosystem.

So, sometimes, in order to:

• improve performances
• mantain a clean data structure
• do DDD right¹

you should go for a NoSQL solution: that said, when you also need to store and retrieve a particular data structure, you should work with a graph database.

Oh My Graph

A graph is a really simple entity ( ok, not that simple: I try to outrageously simplify things here ) with vertices and edges.

Edges connect various vertices, and can be directed or undirected: directed connections happen when the edge has an orientation ( like in the case of a person having a car ), undirected when it hasn’t ( like a person being friend of a person, which is a bi-directional connection ).

You use and work with graphs more frequently than what you can actually think: when dealing with trees ( like the ones you tend to solve with Doctrine’s nested sets ) you are basically working with a directed graph with no cycles.

In the database domain, the graph has an additional element, the property: N properties can be assigned to a vertex ( thus any graph storage engine can work as a document-oriented DB ).

Why OrientDB

OrientDB is a NoSQL graph database with an aim: be easy by making you more productive.

So it’s really easy to install and setup: you just need to download the latest release from googlecode ( hope they will move it to GitHub soon ), unzip it and start the server:

1
2
3
4

cd path/to/orient
cd bin
chmod +x server.sh
./server.sh

Orient is also SQL-friendly, letting you query the database with SQL-like syntax: since everybody knows SQL, you should be able to take confidence with its query language in a few minutes.

It is ACID and can be schema-*; which means that:

• it can be schema-ful, just like a RDBMS
• it can be schema-less, leveraging the power of no-prior knowledge of data’s structure
• it can be schema-mixed, …well, you guessed it!

Another cool thing is that the distribution comes with a pre-defined set of data: so now you are ready to play with graph stuff… but how?

REST in P…DB

I introduce you OrientStudio, the GUI to manage your Orient’s instance.

OrientStudio is a simple tool bundled with Orient’s package, which listens on port 2480 as you start the server, mainly developed in JavaScript, able to let you:

• CRUD objects, classes and clusters
• query the DB
• explore and navigate your graph’s objects

There’s something really cool about OrientStudio: it works, natively, via HTTP, which means that Orient has an HTTP interface to let you manage your graph.

The interface is defined “RESTful”, in the docs: just don’t be finicky and consider it as RESTful as CouchDB ( so definitely not RESTful, but HTTP-loving ).

Orient also works with its binary protocol, which is a looooooot faster than the HTTP one.

OrientDB and PHP

So, what does this have in common with our so-loved scripting language?

We’ll find it out in my next article, where I’ll tell you how to query OrientDb from PHP.