Graph Databases and PHP

Traditional relational databases — ironically — are not that good at the complex relationships some modern applications need.

Multiple joins and complex sub-queries can gradually take a toll on performance.

Graph Databases, on the other hand, are all about relationships. In this talk we will look at using the popular Neo4j graph database with PHP to build efficient relational data for OmNomHub: not your average recipe site.

Transcript

1. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
   Neo4J is
   AWESOME
   

   View Slide

2. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
   

   View Slide

3. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
   WIIIIIIIE \o/
   “Learn the most by
   sharing your knowledge
   with others” - @coderabbi
   

   View Slide

4. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
   This talk is entry-
   level.
   

   View Slide

5. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
   

   View Slide

6. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
   

   View Slide

7. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
   Michelle Sanver
   

   View Slide

8. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
   Michelle Sanver
   Body Text
   President of PHPWomen
   

   View Slide

9. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
   Michelle Sanver
   Accent(s)!?
   

   View Slide

10. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Michelle Sanver
    Body Text
    

    View Slide

11. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    I’m not a drupal
    developer.
    

    View Slide

12. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    How I got into
    Neo4j
    

    View Slide

13. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs \o/
    

    View Slide

14. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs \o/
    Graph theory has
    been studied since
    Leonard Euler’s
    Bridges 1736
    

    View Slide

15. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs \o/
    Body Text
    

    View Slide

16. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs \o/
    

    View Slide

17. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs \o/
    Graphs are really just connected
    data… They are everywhere
    

    View Slide

18. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs \o/
    

    View Slide

19. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs \o/
    It’s modern.
    

    View Slide

20. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs \o/
    Facebook open graph
    

    View Slide

21. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs \o/
    A graph is an easy way to visualise connected data.
    Michelle Graphs
    likes
    

    View Slide

22. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    Relational question:
    Average age of everyone in this list?
    Graph question:
    Who in here is likely going to the next DrupalCon?
    

    View Slide

23. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    Relational databases have tables
    Recipes
    

    View Slide

24. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    Tables have relationships
    Recipes Groups
    

    View Slide

25. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    Which causes a join table…
    Recipes Groups
    RecipeToGroup
    

    View Slide

26. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    You query via the table
    Recipes Groups
    RecipeToGroup
    

    View Slide

27. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    Imagine having *actual* relations
    Pasta group Spaghetti Italian group
    

    View Slide

28. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    Spaghetti Pasta group
    Italian group
    Recipes Groups
    RecipeToGroup
    Spaghetti:Italian
    Spaghetti:Pasta
    Pasta group Spaghetti Italian group
    

    View Slide

29. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    Relational databases have a structure
    

    View Slide

30. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    Neo4j has a flexible schema
    

    View Slide

31. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    A big query.
    The EAV model.
    

    View Slide

32. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    SET @entityid = '3';
    SELECT ea.attribute_id,
    ea.attribute_code, eav.value AS
    'value', 'varchar' AS 'type'
    FROM catalog_category_entity e
    JOIN
    catalog_category_entity_varchar
    eav
    ON e.entity_id = eav.entity_id
    JOIN eav_attribute ea
    ON eav.attribute_id =
    ea.attribute_id
    WHERE e.entity_id = @entityid
    UNION
    SELECT ea.attribute_id,
    ea.attribute_code, eav.value
    AS 'value', 'int' AS 'type'
    FROM
    catalog_category_entity e
    JOIN
    catalog_category_entity_int
    eav
    ON e.entity_id = eav.entity_id
    JOIN eav_attribute ea
    ON eav.attribute_id =
    ea.attribute_id
    WHERE e.entity_id = @entityid
    UNION
    

    View Slide

33. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    SELECT ea.attribute_id,
    ea.attribute_code, eav.value AS 'value',
    'decimal' AS 'type'
    FROM catalog_category_entity e
    JOIN catalog_category_entity_decimal
    eav
    ON e.entity_id = eav.entity_id
    JOIN eav_attribute ea
    ON eav.attribute_id = ea.attribute_id
    WHERE e.entity_id = @entityid
    UNION
    selecting ea.attribute_id, ea.attribute_code, eav.value AS 'value', 'datetim
    'type'
    FROM catalog_category_entity e
    JOIN catalog_category_entity_datetime eav
    ON e.entity_id = eav.entity_id
    JOIN eav_attribute ea
    ON eav.attribute_id = ea.attribute_id
    WHERE e.entity_id = @entityid
    UNION
    SELECT ea.attribute_id, ea.attribute_code, eav.value AS 'value', 'text' AS
    FROM catalog_category_entity e
    JOIN catalog_category_entity_text eav
    ON e.entity_id = eav.entity_id
    JOIN eav_attribute ea
    ON eav.attribute_id = ea.attribute_id
    WHERE e.entity_id = @entityid
    

    View Slide

34. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    Having a flexible database
    with a schema costs a lot.
    

    View Slide

35. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    For modern complex data…
    

    View Slide

36. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs vs. Relational databases
    Would you still pick a relational
    structure?
    

    View Slide

37. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    neo4j.com
    

    View Slide

38. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Java Based
    

    View Slide

39. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Open Source!
    github.com/neo4j
    

    View Slide

40. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Meetups everywhere
    neo4j.meetup.com
    

    View Slide

41. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs and Neo4j
    Graphs have…
    

    View Slide

42. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs and Neo4j
    ( Nodes )
    

    View Slide

43. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs and Neo4j
    (Node) { Properties }
    As many as you want
    Name:
    Michelle
    Nick:
    geekie
    Name:
    Cees-Jan
    Nick:
    WyriHaximus
    

    View Slide

44. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs and Neo4j
    Nick:
    geekie
    Nick:
    WyriHaximus
    knows
    (Node) [ Relationships ]
    As many as you want
    

    View Slide

45. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs and Neo4j
    Labels
    Nick:
    geekie
    Label: Person
    

    View Slide

46. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs and Neo4j
    Indexes for easy lookup
    

    View Slide

47. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs and Neo4j
    Common named graphs
    

    View Slide

48. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs and Neo4j
    Common named graphs
    Local McLaughlin graph
    

    View Slide

49. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs and Neo4j
    Common named graphs
    Diamond Butterfly Bull
    Franklin Tutte Wagner
    

    View Slide

50. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Graphs and Neo4j
    You can make art out of your DB.
    (Don’t, or do?)
    

    View Slide

51. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    OmNomHub
    An idea for a different
    recipe website
    

    View Slide

52. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    OmNomHub
    Like GitHub but for recipes!
    

    View Slide

53. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    OmNomHub
    Fork a recipe
    

    View Slide

54. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    OmNomHub
    See all forks
    

    View Slide

55. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    OmNomHub
    Send change requests
    

    View Slide

56. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    OmNomHub
    Join groups and share recipe
    collections among them
    

    View Slide

57. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    OmNomHub
    Have your own private recipe
    collections
    

    View Slide

58. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    OmNomHub
    Search similar recipes
    (This is a part where the graph structure shines)
    

    View Slide

59. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    OmNomHub
    Honestly… It’s a pet project
    

    View Slide

60. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j architecture
    

    View Slide

61. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j architecture
    Disks
    File System Cache
    Record Files
    Object Cache
    Cypher
    Core API
    Traversal API
    Transaction
    Management
    Transaction Log
    

    View Slide

62. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j architecture
    Disks
    File System Cache
    Record Files
    Object Cache
    Cypher
    Core API
    Traversal API
    Transaction
    Management
    Transaction Log
    

    View Slide

63. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j architecture
    Disks
    File System Cache
    Record Files
    Object Cache
    Cypher
    Core API
    Traversal API
    Transaction
    Management
    Transaction Log
    

    View Slide

64. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j architecture
    Disks
    File System Cache
    Record Files
    Object Cache
    Cypher
    Core API
    Traversal API
    Transaction
    Management
    Transaction Log
    

    View Slide

65. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j architecture
    

    View Slide

66. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j architecture
    Disks
    File System Cache
    Record Files
    Object Cache
    Cypher
    Core API
    Traversal API
    Transaction
    Management
    Transaction Log
    

    View Slide

67. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Cypher and
    Browser
    

    View Slide

68. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Cypher
    Cypher: SQL for graph databases.
    

    View Slide

69. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Cypher
    )
    Parentheses means nodes
    (Or hugs)
    (
    

    View Slide

70. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Cypher
    }
    Curly braces means
    properties
    {
    

    View Slide

71. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Cypher
    ]
    Square brackets means
    relationships
    [
    

    View Slide

72. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Browser
    Makes it easy to visualise and query
    the data.
    

    View Slide

73. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Browser
    Let’s learn cypher in the browser!
    

    View Slide

74. DEMO: Cypher and Browser
    (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    

    View Slide

75. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j architecture
    Neostore File Storage
    

    View Slide

76. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neostore
    Several different store files
    

    View Slide

77. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neostore
    Each store has specific data
    

    View Slide

78. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neostore
    Nodes
    neostore.nodestore.db
    
    9 bytes
    

    View Slide

79. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neostore
    Fixed size == FAST
    

    View Slide

80. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neostore
    Relationships
    neostore.relationshipstore.
    db
    
    33 bytes
    

    View Slide

81. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neostore
    Properties
    neostore.propertystore.db
    
    neostore.propertystore.db.
    index
    
    neostore.propertystore.db.
    strings
    
    neostore.propertystore.db.
    arrays
    

    View Slide

82. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Hardware
    matters!
    

    View Slide

83. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Caching in memory,
    or filesystem
    

    View Slide

84. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    

    View Slide

85. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    There is an actively maintained
    Drupal module for Drupal 7 and 8.
    https://www.drupal.org/project/neo4j
    

    View Slide

86. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    As we saw
    There’s a REST interface!
    

    View Slide

87. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    GraphAware PHP Client
    https://github.com/graphaware/neo4j-php-client
    

    View Slide

88. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    $client = ClientBuilder::create()

    ->addConnection('bolt', 'bolt://neo4j:password@localhost:7687')

    ->build();
    

    $query = "MATCH (n:User) RETURN n";

    $result = $client->run($query);

    

    foreach ($result->getRecords() as $record) {

    echo(count($record->value('name'));

    }

    

    View Slide

89. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    $client = ClientBuilder::create()

    ->addConnection('bolt', 'bolt://neo4j:password@localhost:7687')

    ->build();
    

    $query = "MATCH (n:User) RETURN n";

    $result = $client->run($query);

    

    foreach ($result->getRecords() as $record) {

    echo(count($record->value('name'));

    }

    

    View Slide

90. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    $client = ClientBuilder::create()

    ->addConnection('bolt', 'bolt://neo4j:password@localhost:7687')

    ->build();
    

    $query = "MATCH (n:User) RETURN n";

    $result = $client->run($query);

    

    foreach ($result->getRecords() as $record) {

    echo(count($record->value('name'));

    }

    

    View Slide

91. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    $client = ClientBuilder::create()

    ->addConnection('bolt', 'bolt://neo4j:password@localhost:7687')

    ->build();
    

    $query = "MATCH (n:User) RETURN n";

    $result = $client->run($query);

    

    foreach ($result->getRecords() as $record) {

    echo(count($record->value('name'));

    }

    

    View Slide

92. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    GraphAware also released an OGM
    

    View Slide

93. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    /**

    * @OGM\GraphId()

    * @var int

    */

    protected $id;

    

    /**

    * @OGM\Property(type="string")

    * @var string

    */

    protected $recipeName;

    

    /**

    * @OGM\Relationship(type="FORKED", direction="OUTGOING",
    targetEntity="User", collection=true)

    * @var ArrayCollection|User[]

    */

    protected $users;
    

    View Slide

94. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    /**

    * @OGM\GraphId()

    * @var int

    */

    protected $id;

    

    /**

    * @OGM\Property(type="string")

    * @var string

    */

    protected $recipeName;

    

    /**

    * @OGM\Relationship(type="FORKED", direction="OUTGOING",
    targetEntity="User", collection=true)

    * @var ArrayCollection|User[]

    */

    protected $users;
    

    View Slide

95. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    /**

    * @OGM\GraphId()

    * @var int

    */

    protected $id;

    

    /**

    * @OGM\Property(type="string")

    * @var string

    */

    protected $recipeName;

    

    /**

    * @OGM\Relationship(type="FORKED", direction="OUTGOING",
    targetEntity="User", collection=true)

    * @var ArrayCollection|User[]

    */

    protected $users;
    

    View Slide

96. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    /**

    * @OGM\GraphId()

    * @var int

    */

    protected $id;

    

    /**

    * @OGM\Property(type="string")

    * @var string

    */

    protected $recipeName;

    

    /**

    * @OGM\Relationship(type="FORKED", direction="OUTGOING",
    targetEntity="User", collection=true)

    * @var ArrayCollection|User[]

    */

    protected $users;
    

    View Slide

97. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    $user = $em->getRepository(User::class)-
    >findOneBy('name', 'Michelle');
    

    View Slide

98. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    Neo4j in PHP
    It is well documented
    https://neo4j.com/developer/php/
    https://github.com/graphaware/neo4j-php-client
    https://github.com/graphaware/neo4j-php-ogm
    

    View Slide

99. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
    OmNomHub
    The possible Future
    

    View Slide

100. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     OmNomHub: The Future
     Connecting users
     “You both like”
     

     View Slide

101. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     OmNomHub: The Future
     Connecting recipes
     “These have similar ingredients and
     user base”
     

     View Slide

102. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     OmNomHub: The Future
     Being smart
     “You might like”
     

     View Slide

103. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     OmNomHub: The Future
     Being smart
     Smart recipe collections!
     

     View Slide

104. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     OmNomHub: The Future
     Being creepy
     “Don’t like meat huh?”
     

     View Slide

105. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     Wrapup
     

     View Slide

106. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     Wrapup
     Graphs are everywhere
     

     View Slide

107. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     Wrapup
     They make it easy to connect data
     

     View Slide

108. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     Wrapup
     Easier to be creepy <3
     

     View Slide

109. Resources
     (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     neo4j.org/learn
     https://neo4j.com/developer/php/
     https://www.drupal.org/project/neo4j
     

     View Slide

110. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     Thank you!
     

     View Slide

111. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     Questions?
     

     View Slide

112. (Michelle)-[:LOVES]->(Neo4j) @michellesanver
     How would you
     use it?
     

     View Slide