Loading

Graph Training 1

XN Logic Corporation

This is a live streamed presentation. You will automatically follow the presenter and see the slide they're currently on.

Introduction to Graph Databases & The Pacer library

Brought to you by

What Is a Graph?

Vertices 

(aka nodes)

Edges 

(aka relations)

  • A data structure
  • Used for storing and accessing data

+

Variations

Undirected

Directed

vs.

Variations

Without

With Vertex Properties

vs.

Variations

Without

With Edge Labels

vs.

Property Graph

  • Directed graph

 

  • Vertices & Edges with properties
    • Key-value pairs
    • Keys are strings, values can be anything

 

  • Edges are labeled
    • Each edge has exactly one label (a string).
    • Label can be thought of as "edge type".

Property Graph

Conceptually simple,

yet extremely flexible ... 

Airline's Flight Data

Airline's Flight Data

Social Network

Social Network

Decision Making

One way of representing the data in a property graph ... 

Decision Making

IT Infrastructure

IT Infrastructure

Graph Databases

So far, we've seen an abstract model.

In practice, we will need a graph database.

A graph database is software that:

  • Stores a property graph
  • Supports the basic graph operations
    • Given a vertex, get its edges
    • Given an edge, get its vertices
  • Does it efficiently!

Graph Databases

There are a number of graph databases out there:

Graph Databases

Databases differ in many aspects:

  • Performance
  • Persistence strategies
  • Transaction management
  • Supported data types
  • Features
    • High-Availability
    • Clustering
  • Integrations
    • Indexing/searching (e.g. Lucene)
    • Caching (e.g. Memcached)
  • Licensing & Price

Choosing a Graph DB

  • Your specific needs (e.g. Scalability, flexibility, ease of development and/or deployment, etc).
  • Existing ecosystem.
  • How difficult is it to switch to a different database?
  • Can I use multiple databases together?

Things to consider ...

Which graph database is the best?

Depends on your use-case.

Luckily, we don't have to commit to one database...

Blueprints

  • Blueprints
  • ​Generic Java interfaces for property graphs
  • Adapters for all common databases​
    • Build your application against Blueprints, and it will support all common graph databases.
  • ​Open-source
  • Supported by XN Logic

Additional Benefits

  • Build Blueprints adapters for other data stores, and use them as if they were graph databases.
  • Use other libraries from the Tinkerpop stack.
  • Use Pacer:
    • Efficient graph traversal
    • Extensible data modeling
    • Open-source

Pacer

  • Written in JRuby
    • Ruby expressiveness
    • Java performance 
  • Built on top of the Tinkerpop stack
  • Well-tested
  • Used in production

If you haven't done so already, you may want to install Pacer and its dependencies.

Pacer in 60 Seconds

Graph traversal

  • Graphs are processed as streams of elements
  • Routes = composable, lazily-loaded streams.
  • Build traversals by chaining routes

Data Modeling

  • Extend elements with Ruby modules
  • Use extensions to create a Domain Specific Language

The Benefits

  1. Routes allow us to run queries efficiently.
  2. DSL makes it easy to define complex queries.
user.as(:me)
    .out_e(:friend).in_v(type: "user").as(:my_friends)
    .out_e(:friend).in_v(type: "user")
    .is_not(:me).is_not(:my_friends)

becomes ...

user.suggest_friends

OK, let's see Pacer in action ...

Made with Slides.com