Welcome to the matrix !
With Neo4j
Neo4j is a NO-SQL Graph database, it's written in java and it's open-source* !* But not only
NOSQL
Big & Linked data
Some limitation
- 2ˆ35 (~ 34 billion) nodes
- 2ˆ35 (~ 34 billion) relationships
- 2ˆ36 (~ 68 billion) properties
- 2ˆ15 (~ 32 000) relationship types
What is a graph database ?
Find out if a person get a specific role
Find out the possible new friends based on them liking similar things
How to query my graph ?
Indexes
Traversal
Client
- REST API
- Java : natif client (embedded), spring-data, play!
- Ruby, Python, Php, Javascript, .Net ...
REST API - Node
http://localhost:7474/db/data/node/{id}
http://localhost:7474/db/data/node/{id}/properties/{key}
http://localhost:7474/db/data/node/{id}/relationships/{in|out}
-
curl --header "Content-Type: application/json" --data '{"foo":"bar"}' http://localhost:7474/db/data/node
-
curl http://localhost:7474/db/data/node/14
REST API - Relationship
http://localhost:7474/db/data/relationship/{id}
http://localhost:7474/db/data/relationship/{id}/properties/{key}
http://localhost:7474/db/data/node/{id}/relationships
-
curl --header "Content-Type: application/json" --data '{"to" : "http://localhost:7474/db/data/node/6", "type" : "LINK", "data" : {"foo" : "bar"}}' http://localhost:7474/db/data/node/0/relationships
-
curl http://localhost:7474/db/data/relationship/25
Some DSL
- Write in Groovy
- TinkerGraph, Neo4j, OrientDB, DEX, InfiniteGraph, Rexster, and Sail RDF Stores
- Graphe (G) = Vertex(node) , Edge(relationship)
- Everything in Gremlin is an iterator
- Each operation is a pipe
// Display "name" of all node
g.V.name
// Complexe query :
g.v(23).out('knows').filter{it.age > 30}.out('like').name
Cypher
- Human query language
- Inspired by SQL
- Only avaible for Neo4j
Cypher - Pattern
- START: Starting points in the graph, obtained by element IDs or via index lookups
-
MATCH: The graph pattern to match, bound to the starting points in
START - WHERE: Filtering criteria
- RETURN: What to return
Cypher - sample query
// All node related to node 3
start n=(3) match (n)--(x) return x
// All node that are BLOCKed by node 3
start n=(3) match (n)-[r, :BLOCKS]-> (x) return x
// All BLOCKS relationships outgoing from node 3
start n=(3) match (n)-[r, :BLOCKS]-> () return r
// Where regex
start n=(2,1) match n.name =~ /Tob.*/ return n
Cypher - complexe query
START
joe=node:node_auto_index(name = "Joe")
MATCH joe-[:knows]->friend-[:knows]->friend_of_friend,
joe-[r?:knows]->friend_of_friend
WHERE r IS NULL
RETURN friend_of_friend.name, COUNT(*)
ORDER BY COUNT (*) DESC, friend_of_friend.name
Transaction
All write operations must be performed in a transaction.
Cycle
- Begin a transaction
- Operate on the graph performing write operations
- Mark the transaction as successful or not
- Finish the transaction
Lock
Lock
// Working on a property of a node (or a relationship)
lock the node (or the relationship)
// Create, remove a node
lock the node
// Create, remove a relationship
lock relationship and its nodes
In memory write operation
How to extend Neo4j ?
Extension
Don't care of endpoint,JSON format, It's a part of the Neo4j process !
Plugin
Based on JAX-RS, You can do what ever you want, JSON result depend on you !
List of extension
- Gremlin & Cypher
- RDF / Sparql
- Spatial
Coming soon routing with OSM data
Spatial & OSM
Backup
- Save the index !
- Hot backup process (EE)
Cluster
Master
- Unique
- Read / write
- Elected by the cluster (Via Apache ZooKeeper)
Slaves
- Many
- Read-only
- redirect write query to master
Thanks to
- Neo-technology !
- Peter Neubauer (I re-use some illustration)
- To the neo4j community
- impress.js (it's not a prezi presentation !)
Use a spacebar or arrow keys to navigate