This is the presentation given by Michael Hunger and Peter Neubauer at the SF Data Mining group, see http://www.meetup.com/Data-Mining/events/80275492/

1. Neo4j -
from 0 to Her0
Thank you Trulia!
Peter Neubauer Andreas Kollegger Michael Hunger
@peterneubauer @akollegger @mesirii
#neo4j
1

2. Neo4j -
from 0 to Her0
Thank you Trulia!
Peter Neubauer Andreas Kollegger Michael Hunger
@peterneubauer @akollegger @mesirii
#neo4j
1

3. Neo4j
Neo4j -
from 0 to Her0
Thank you Trulia!
Peter Neubauer Andreas Kollegger Michael Hunger
@peterneubauer @akollegger @mesirii
#neo4j
1

4. 1

5. 2

6. Um, Neo for what?
2

7. Um, Neo for what?
•Neo4j - the graph database
•Graph
•no: not for charts & diagrams, or vector artwork
•yes: for storing data that is connected
•remember linker lists, trees?
•graphs are the general-purpose data structure
2

8. Neo4j is a Graph Database
3

9. Neo4j is a Graph Database
๏ A Graph Database:
3

10. Neo4j is a Graph Database
๏ A Graph Database:
• a Property Graph with Nodes, Relationships
and Properties on both
3

11. Neo4j is a Graph Database
๏ A Graph Database:
• a Property Graph with Nodes, Relationships
and Properties on both
• perfect for complex, highly connected data
3

12. Neo4j is a Graph Database
๏ A Graph Database:
• a Property Graph with Nodes, Relationships
and Properties on both
• perfect for complex, highly connected data
๏ A Graph Database:
3

13. Neo4j is a Graph Database
๏ A Graph Database:
• a Property Graph with Nodes, Relationships
and Properties on both
• perfect for complex, highly connected data
๏ A Graph Database:
• reliable with real ACID Transactions
3

14. Neo4j is a Graph Database
๏ A Graph Database:
• a Property Graph with Nodes, Relationships
and Properties on both
• perfect for complex, highly connected data
๏ A Graph Database:
• reliable with real ACID Transactions
• scalable: 32 Billion Nodes, 32 Billion Relationships, 64 Billion
Properties
3

15. Neo4j is a Graph Database
๏ A Graph Database:
• a Property Graph with Nodes, Relationships
and Properties on both
• perfect for complex, highly connected data
๏ A Graph Database:
• reliable with real ACID Transactions
• scalable: 32 Billion Nodes, 32 Billion Relationships, 64 Billion
Properties
• Server with REST API, or Embeddable on the JVM
3

16. Neo4j is a Graph Database
๏ A Graph Database:
• a Property Graph with Nodes, Relationships
and Properties on both
• perfect for complex, highly connected data
๏ A Graph Database:
• reliable with real ACID Transactions
• scalable: 32 Billion Nodes, 32 Billion Relationships, 64 Billion
Properties
• Server with REST API, or Embeddable on the JVM
• high-performance with High-Availability (read scaling) 3

17. Graph DB 101
4

18. 5

19. You know relational
5

20. You know relational
5

21. You know relational
foo
5

22. You know relational
foo bar
5

23. You know relational
foo foo_bar bar
5

24. You know relational
foo foo_bar bar
5

25. You know relational
foo foo_bar bar
5

26. You know relational
foo foo_bar bar
5

27. You know relational
now consider relationships...
5

28. You know relational
now consider relationships...
5

29. You know relational
now consider relationships...
5

30. You know relational
now consider relationships...
5

31. You know relational
now consider relationships...
5

32. You know relational
now consider relationships...
5

33. 5

34. 6

35. We're talking about a
Property Graph
6

36. We're talking about a
Property Graph
Nodes
6

37. We're talking about a
Property Graph
Nodes
Relationships
6

38. We're talking about a
Property Graph
Em Joh
il a n
knows knows
Alli Tob Lar
Nodes
son ias knows s
knows
And And knows
knows rea rés
s
knows knows knows
Pet Miic
Mc knows Ian
er knows a
a
knows knows
De Mic
lia h ael
Relationships
Properties (each a key+value)
+ Indexes (for easy look-ups)
6

39. 6

40. 7

41. And, but, so how do you
query this "graph" database?
7

42. Cypher Query Language

43. What is Cypher?
• Graph Query Language for Neo4j
• Aims to make querying simple

44. Design Decisions
Pattern matching

45. Design Decisions
Pattern matching
A
B C

46. Design Decisions
Pattern matching

47. Design Decisions
Pattern matching

48. Design Decisions
Pattern matching

49. Design Decisions
Pattern matching

50. Design Decisions
ASCII-art patterns

51. Design Decisions
ASCII-art patterns
() --> ()

52. Design Decisions
Directed relationship
A B

53. Design Decisions
Directed relationship
A B
(A) --> (B)

54. Design Decisions
Undirected relationship
A B

55. Design Decisions
Undirected relationship
A B
(A) -- (B)

56. Design Decisions
specific relationships
LOVES
A B

57. Design Decisions
specific relationships
LOVES
A B
A -[:LOVES]-> B

58. Design Decisions
Joined paths
A B C

59. Design Decisions
Joined paths
A B C
A --> B --> C

60. Design Decisions
multiple paths
A
B C

61. Design Decisions
multiple paths
A
B C
A --> B --> C, A --> C

62. Design Decisions
multiple paths
A
B C
A --> B --> C, A --> C
A --> B --> C <-- A

63. Design Decisions
Optional relationships
A B

64. Design Decisions
Optional relationships
A B
A -[?]-> B

65. SELECT skills.*, user_skill.*
FROM users
JOIN user_skill ON users.id = user_skill.user_id
JOIN skills ON user_skill.skill_id = skill.id WHERE
users.id = 1

66. START user = node(1)
MATCH user -[user_skill]-> skill
RETURN skill, user_skill

67. Indexes
Used as multiple starting points, not to speed
up any traversals
START a = node:nodes_index(type='User') MATCH
a-[r:knows]-b
RETURN ID(a), ID(b), r.weight

68. Variable length Path Match
Some UGLY recursive self join on the groups
table
START max=node:person(name=“Max")
MATCH group <-[:BELONGS_TO*]- max
RETURN group

69. 27

70. Who uses this stuff?
27

72. [A] Mmm Pancakes

73. [A] Mmm Pancakes

74. Cute meta + data
This Material is subject to the terms of the Mozilla Public # License, v. 2.0. If a copy of the
MPL was not distributed with this # file, You can obtain one at http://mozilla.org/MPL/2.0/

75. Cute meta + data
This Material is subject to the terms of the Mozilla Public # License, v. 2.0. If a copy of the
MPL was not distributed with this # file, You can obtain one at http://mozilla.org/MPL/2.0/

76. Neo4J Co-Existence
• Node uuids as refs in external ElasticSearch
also in internal Lucene
• Custom search ranking for user history based on
node relationship data
• MySQL for user data, Redis for metrics
This Material is subject to the terms of the Mozilla Public # License, v. 2.0. If a copy of the
MPL was not distributed with this # file, You can obtain one at http://mozilla.org/MPL/2.0/

78. [B] ACL from Hell
One of the top 10 telcos worldwide

79. Example Access Authorization
User
Access may be given directly or by inheritance Customer
Account
U Subscription
U Inherit = true
Inherit =
false
In
C
he
rit
=
C C C
tru
e
A A A A A
S S S S S S S S S S

81. [C] Master of your Domain

82. [C] MDM within Cisco
master data management, sales compensation management, online customer support
Description Benefits
Real-time conflict detection in sales compensation management. Performance : “Minutes to Milliseconds”
Business-critical “P1” system. Neo4j allows Cisco to model complex Outperforms Oracle RAC, serving complex queries in real time
algorithms, which still maintaining high performance over a large Flexibility
dataset. Allows for Cisco to model interconnected data and complex queries with
ease
Background
Robustness
Neo4j replaces Oracle RAC, which was not performant enough for the
With 9+ years of production experience, Neo4j brings a solid product.
use case.
Architecture
3-node Enterprise cluster with mirrored
disaster recovery cluster
Dedicated hardware in own datacenter
Embedded in custom webapp
Sizing
35 million nodes
50 million relationships
600 million properties

83. 35

84. Webadmin
35

85. 36

86. Graph -[:Cypher]-> JDBC
36

87. 37

88. Today’s app:
Cineasts
37

89. 38

90. The domain model
38

91. 39

92. Hands-on: Installation
39

93. Hands-on: Installation
•copy zip drive contents to your box
•cd cineasts
•./import.sh
•neo4j/bin/neo4j-shell
•http://localhost:7474
39

94. Some Links
๏ GraphConnect
๏ Community: http://neo4j.org
40

95. 41

96. Cypher - overview
๏ a pattern-matching query language
๏ declarative grammar with clauses (like SQL)
๏ aggregation, ordering, limits
๏ create, read, update, delete
41

97. 42

98. Cypher - read clauses
๏ Find a place to begin:
•START <lookup>
๏ Describe what to find:
•MATCH <paths>
๏ Filter the elements:
•WHERE <filters>
๏ RETURN <values>
42

99. 43

100. Cypher: START + RETURN
๏ START <lookup> RETURN <expressions>
๏ START binds terms using simple look-up
•directly using known ids
•or based on indexed Property
๏ RETURN expressions specify result set
43

101. Cypher: START + RETURN
๏ START <lookup> RETURN <expressions>
๏ START binds terms using simple look-up
•directly using known ids
•or based on indexed Property
๏ RETURN expressions specify result set
// lookup node id 1, return that node
start n=node(1) return n
// lookup all nodes
start n=node(*) return n
43

102. 44

103. Cypher: MATCH
๏ START <lookup> MATCH <pattern> RETURN <expr>
๏ MATCH describes a pattern of nodes+relationships
•node terms in optional parenthesis
•lines with arrows for relationships
44

104. Cypher: MATCH
๏ START <lookup> MATCH <pattern> RETURN <expr>
๏ MATCH describes a pattern of nodes+relationships
•node terms in optional parenthesis
•lines with arrows for relationships
// lookup 'n', traverse any relationship to some 'm'
start n=node(0) match (n)--(m) return n,m;
// any outgoing relationship from 'n' to 'm'
start n=node(1) match n-->m return n,m;
// only 'next' relationships from 'n' to 'm' up to 3 away
start n=node(1) match p=n-[:next*..3]->m return p;
// from 'n' to 'm' and capture the relationship as 'r'
start n=node(1) match n-[r]->m return n,r,m
44

105. 45

106. Cypher: WHERE
๏ START <lookup> [MATCH <pattern>]
WHERE <condition> RETURN <expr>
๏ WHERE filters nodes or relationships
•uses expressions to constrain elements
45

107. Cypher: WHERE
๏ START <lookup> [MATCH <pattern>]
WHERE <condition> RETURN <expr>
๏ WHERE filters nodes or relationships
•uses expressions to constrain elements
// lookup all nodes as 'n', constrained to name 'Andreas'
start n=node(*) where n.name='Andreas' return n
// filter nodes where age is less than 30
start n=node(*) where n.age<30 return n
// filter using a regular expression
start n=node(*) where n.name =~ ‘Tob.*’ return n
// filter for a property exists
start n=node(*) where has(n.name) return n
45

108. 46

109. Cypher: CREATE
๏ CREATE <node>[,node or relationship] RETURN <expr>
•create nodes with optional properties
•create relationship (must have a type)
46

110. Cypher: CREATE
๏ CREATE <node>[,node or relationship] RETURN <expr>
•create nodes with optional properties
•create relationship (must have a type)
// create an anonymous node
create n
// create node with a property, returning it
create n={name:'Andreas'} return n
// lookup 2 nodes, then create a relationship and return it
start n=node(0),m=node(1) create n-[r:KNOWS]-m return r
// lookup nodes, then create a relationship with properties
start n=node(1),m=node(2) create n-[r:KNOWS {since:2008}]->m
46

111. 47

112. Cypher: SET
๏ SET [<node property>] [<relationship property>]
•update a property on a node or relationship
•must follow a START
47

113. Cypher: SET
๏ SET [<node property>] [<relationship property>]
•update a property on a node or relationship
•must follow a START
// update the name property
start n=node(0) set n.name='Peter'
// update many nodes, using a calculation
start n=node(*) set n.size=n.size+1
// match & capture a relationship, update a property
start n=node(1) match n-[r]-m set r.times=10
47

114. 48

115. Cypher: DELETE
๏ DELETE [<node>|<relationship>|<property>]
•delete a node, relationship or property
•must follow a START
•tofirst a node, all relationships must be deleted
delete
48

116. Cypher: DELETE
๏ DELETE [<node>|<relationship>|<property>]
•delete a node, relationship or property
•must follow a START
•tofirst a node, all relationships must be deleted
delete
// delete a node
start n=node(5) delete n
// remove a node and all relationships
start n=node(3) match n-[r]-() delete n, r
// remove a property
start n=node(3) delete n.age
48