Intro to Graph Databases

and Neo4j

About Me

@dabernathy89

What's a graph?

Nodes

And relationships between nodes

Property Graph Model

​

http://neo4j.com/developer/graph-database/#property-graph

Animal

"name" : "Banana",
"color" : "orange"

Furnishing

"type" : "carpet"

Scratched

"timestamp" : 1417126199

Barfed On

"timestamp" : 1417136235

  • Nodes and relationships have properties
  • Relationships have names and a direction
  • Any number of relationships between nodes

Relationships

  • Relationships are considered "first class citizens"
    • just as important as nodes
  • Allows for "index free adjacency"
    • nodes point directly to connected nodes
  • Even with many 'joins', query performance is only limited by the portion of the graph that is searched.

Graph: E-Commerce

Stored the same way you might describe your data naturally

 

Reflects your application's domain

NoSQL

Graph databases are part of the NoSQL family

... but they're pretty different.

NoSQL

Aggregate-Oriented

  • Key/Value
  • Column Family
  • Document Store

2 main categories of NoSQL databases:

Graph Databases

When people talk about NoSQL, they're usually referring to the aggregate oriented databases.

reference

Querying with Cypher

Parentheses = nodes

(user)

Arrows and square brackets = relationships

-[:has]->

Together

(user)-[:has]->(role)

Querying with Cypher

Variables

(foo:User)-[bar:has]->()
  • Used for referring to later in the query

Labels

  • let you differentiate between nodes
  • multiple are allowed
(creature:Dog)

Relationship types

  • required, only one allowed
[:purchased]

Querying with Cypher

Properties

 

(:Customer {name:"Daniel"})
-[:PURCHASED {timestamp: 1417631772}]->
(:Product {id: "abc", price: 35.5})

Querying with Cypher

Querying is about matching patterns

 

Example: find customers who have purchased the same product as current customer

MATCH (c:Customer)-[:PURCHASED]->()<-[:PURCHASED]-(c2)
WHERE c.email = "customer@example.com"
return c2

Performance

This graph search...

... has the same performance as this one

Graph advantages summarized:

  • Graph databases are a natural representation of your data.
  • Graph query languages like Cypher are extremely readable.
  • Graphs make complex questions simpler to ask.
  • When dealing with many relationships and complex joins, graph databases can provide better performance.
  • They're fun, especially with visualizations.

PHP

Resources

Videos

References / Training

Books

  • "Graph Databases" - O'Reilly
  • "Learning Neo4J" - Packt

Services

PHP/Neo4j People

Made with Slides.com