Graph the Planet!

POSSCON 2015/04/14

Many developers have become familiar with relational or document based database systems. Relational databases give you the promise of flexible querying at the cost of larger upfront storage design and potential for the “join bomb”. Document databases give the promise of flexible storage and quick, simple retrieval at the cost of retrieval design. Both have proven themselves in application development but they have also proven there is a pitfall, highly connected data presents many challenges. Enter graph databases, whose love of relationships could fill the gap.

Graph databases have proven to be a powerful force in logistical planning, music recommendations, static code analytics, but why? What is it about this data structure that makes it so good at modeling real world problems? Join me on a journey down through nodes and relationships to discover what differentiates graph databases from others, what pitfalls to avoid when building and querying a graph, and unlock the power of graph databases with a series of examples built on top of Neo4j, the leading engine for your graph storage needs.

Transcript

1. GRAPH THE PLANET! Introduction to the world of circles and arrows
2. WHO ARE YOU AGAIN? Sean Copenhaver GIS Columbia, SC Twitter: @copenhas Github: copenhas VainGlory: copenhas
3. WHAT IS A GRAPH?
4. NODES/VERTICES
5. NODES/VERTICES
6. RELATIONSHIPS/EDGES
7. RELATIONSHIPS/EDGES
8. A GRAPH
9. WHAT DO YOU DO WITH IT?
10. TRAVERSE IT
11. QUICK ASIDE TO RDBMS Artist Album User UserLikes
12. WHAT BANDS WOULD COPENHAS LIKE?
13. LET’S TRAVERSE THE TABLES select Artist.Name from User join UserLikes on User.Id = UserLikes.UserId join Album on UserLikes.AlbumId = Album.Id join Artist on Album.ArtistId = Artist.Id where User.Name = 'copenhas'
14. JOIN BOMB
15. CAN A GRAPH DB HELP?
16. LET’S TRAVERSE THE TABLES AGAIN match (user:User)-[:LIKES]->(album:Album) match album<-[:RELEASED]-(artist:Artist) where user.Name = ‘copenhas' return artist.Name
17. MAYBE USE A DOC DB? { "name": "Blind Guardian", "origin": "Germany", "genre": "Power Metal", "releases": [ { "name": "Nightfall in Middle-Earth", "year": 1998, "label": "Virgin" } ] }
18. MAYBE USE A DOC DB? Joins are due to normalization You wouldn’t miss that normalization Data has few and shallow relationships
19. BENEFITS OF GRAPH DATABASES Whiteboard friendly Just as flexible as a document DB (almost) “Index-free adjacency” Graphs love relationships
20. NEO4J
21. IT’S GOT WHAT YOU NEED ACID Transactions Schemas (limited), indexes Built in graph algorithms Query language (Cypher)
22. HIGH AVAILABILITY Montag, David. (2013). Understanding Neo4j Scalability. Available: http://info.neo4j.com/rs/neotechnology/images/Understanding%20Neo4j%20Scalability%282%29.pdf. Last accessed 10th April 2015.
23. NOT TO MENTION NEO4J BROWSER!