Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Riak -- Seattle Scalability

Riak -- Seattle Scalability

Presentation I about Riak/future-of-Riak at the Seattle Scalability meetup in August 2013. Note: slides are designed for presentation and miss context without me talking.

Joseph Blomstedt

August 28, 2013
Tweet

More Decks by Joseph Blomstedt

Other Decks in Programming

Transcript

  1. Joseph Blomstedt (@jtuple)
    Basho Technologies
    Seattle Scalability Meetup
    August 2013

    View Slide

  2. 2
    Joe Blomstedt

    View Slide

  3. 3
    Senior Engineer

    View Slide

  4. View Slide

  5. 5
    @jtuple

    View Slide

  6. 6

    View Slide

  7. 7

    View Slide

  8. 8
    Open Source

    View Slide

  9. 9
    Open Source
    Apache 2
    github.com/basho

    View Slide

  10. 10

    View Slide

  11. 11
    Multi-DC Replication
    Support

    View Slide

  12. Riak Basics
    12

    View Slide

  13. 13

    View Slide

  14. 14
    Key / Value
    Secondary Indexes
    Full Text Search
    Map/Reduce
    Counters

    View Slide

  15. 15
    Key / Value
    Secondary Indexes
    Full Text Search
    Map/Reduce
    Counters

    View Slide

  16. 16
    PUT
    GET
    DELETE

    View Slide

  17. 17
    users/joe
    PUT

    View Slide

  18. 18
    users/joe
    GET

    View Slide

  19. 19
    users/joe
    DELETE

    View Slide

  20. 20
    Key / Value
    Secondary Indexes
    Full Text Search
    Map/Reduce
    Counters

    View Slide

  21. 21
    user/1234
    PUT
    {first: “John”,
    last: “Doe”,
    age: 40}

    View Slide

  22. 22
    user/1234
    PUT
    {first: “John”,
    last: “Doe”,
    age: 40}
    first_bin: “John”
    last_bin: “Doe”
    age_int: 40

    View Slide

  23. 23
    pics/pic42
    PUT
    animal_bin: cat
    cuteness_int: 10

    View Slide

  24. 24
    Tagged Fields
    Exact/Range
    Single Index
    Streaming
    Pagination

    View Slide

  25. 25
    Key / Value
    Secondary Indexes
    Full Text Search
    Map/Reduce
    Counters

    View Slide

  26. 26
    Automatic Indexing
    JSON
    XML
    Text
    Custom

    View Slide

  27. 27
    Querying
    Exact Match
    Wildcard
    Range

    View Slide

  28. 28
    Querying
    Grouping
    Prefix matching
    Term Boosting

    View Slide

  29. 29
    Querying
    AND
    OR
    NOT

    View Slide

  30. 30
    Key / Value
    Secondary Indexes
    Full Text Search
    Map/Reduce
    Counters

    View Slide

  31. 31
    Use Cases
    Filtering
    Aggregation
    Extraction

    View Slide

  32. 32
    M/R Input
    Known Keys
    2nd Index Query
    Full Text Query

    View Slide

  33. 33
    Distributed
    Mappers: data nodes
    Reducers: coordinator

    View Slide

  34. 34
    Key / Value
    Secondary Indexes
    Full Text Search
    Map/Reduce
    Counters

    View Slide

  35. 35
    Why do people
    use Riak?

    View Slide

  36. Big Data
    36

    View Slide

  37. Big Data
    37
    Scalable
    Distributed

    View Slide

  38. Big Data
    38
    5 nodes
    100 nodes

    View Slide

  39. Critical Data
    39

    View Slide

  40. Critical Data
    40
    Big
    Medium
    small

    View Slide

  41. Fault Tolerance
    41

    View Slide

  42. Availability
    42

    View Slide

  43. CAP Theorem
    43

    View Slide

  44. 44
    Partition-tolerance
    Consistency
    Availability

    View Slide

  45. 45
    Partition-tolerance
    Consistency Availability

    View Slide

  46. 46
    Partition-tolerance
    Consistency Availability
    CP AP

    View Slide

  47. 47
    Partition-tolerance
    Consistency Availability
    CP AP

    View Slide

  48. 48
    C/P
    Strict Quorum A/P
    Sloppy Quorum A/P

    View Slide

  49. 49
    C/P
    Strict Quorum A/P
    Sloppy Quorum A/P

    View Slide

  50. 50
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  51. 51
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  52. 52
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  53. 53
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  54. 54
    C/P
    Strict Quorum A/P
    Sloppy Quorum A/P

    View Slide

  55. 55
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  56. 56
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  57. 57
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  58. 58
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  59. 59
    C/P
    Strict Quorum A/P
    Sloppy Quorum A/P

    View Slide

  60. 60
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  61. 61
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  62. 62
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  63. 63
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client

    View Slide

  64. 64
    Node 1 Node 2 Node 3 Node 4 Node 5
    client client
    client client client

    View Slide

  65. Eventual Consistency
    65

    View Slide

  66. 66
    A A A

    View Slide

  67. 67
    A A A

    View Slide

  68. 68
    A A A
    B

    View Slide

  69. 69
    A A A
    B

    View Slide

  70. 70
    A A A
    B
    B B B

    View Slide

  71. 71
    A A A

    View Slide

  72. 72
    A A A
    B C

    View Slide

  73. 73
    A A A
    B C

    View Slide

  74. 74
    A A A
    B
    {B,C} {B,C} {B,C}
    C

    View Slide

  75. 75
    Write Once
    Immutable
    Last Write Wins
    Business Rules
    CRDTs/Monotonicity

    View Slide

  76. 76
    Write Once
    Immutable
    Last Write Wins
    Business Rules
    CRDTs/Monotonicity

    View Slide

  77. 77
    Write Once
    Immutable
    Last Write Wins
    Business Rules
    CRDTs/Monotonicity

    View Slide

  78. 78
    Write Once
    Immutable
    Last Write Wins
    Business Rules
    CRDTs/Monotonicity

    View Slide

  79. 79
    Write Once
    Immutable
    Last Write Wins
    Business Rules
    CRDTs/Monotonicity

    View Slide

  80. 80
    Write Once
    Immutable
    Last Write Wins
    Business Rules
    CRDTs/Monotonicity

    View Slide

  81. 81
    Write Once
    Immutable
    Last Write Wins
    Business Rules
    “Set Union”

    View Slide

  82. 82
    {1} {1} {1}

    View Slide

  83. 83
    {1} {1} {1}

    View Slide

  84. 84
    {1} {1} {1}
    {1,2} {1,3} {1,2,4}

    View Slide

  85. 85
    {1} {1} {1}
    {1,2} {1,3} {1,2,4}

    View Slide

  86. 86
    {1} {1} {1}
    {1,2}
    {1,2,3,4} {1,2,3,4} {1,2,3,4}
    {1,3} {1,2,4}

    View Slide

  87. Clustering
    87

    View Slide

  88. 88
    Consistent Hashing
    Sloppy Quorums
    Hinted Handoff

    View Slide

  89. 89
    0 2160

    View Slide

  90. 90
    0
    2160

    View Slide

  91. 91

    View Slide

  92. 92
    Node 2
    Node 1
    Node 3

    View Slide

  93. 93
    Node 2
    Node 1
    Node 3
    Node 4

    View Slide

  94. 94
    Node 2
    Node 1
    Node 3
    Node 4
    SHA1(Key)

    View Slide

  95. 95
    Node 2
    Node 1
    Node 3
    Node 4
    SHA1(Key)
    primary
    replicas

    View Slide

  96. 96
    Node 2
    Node 1
    Node 3
    Node 4
    primary
    +
    fallback
    replicas

    View Slide

  97. 97
    Node 2
    Node 1
    Node 3
    Node 4
    handoff

    View Slide

  98. Self Healing
    98

    View Slide

  99. 99
    Here Today
    !=
    Here Tomorrow

    View Slide

  100. 100
    Read Repair

    View Slide

  101. 101
    Active Anti Entropy

    View Slide

  102. 102
    Node 2
    Node 1
    Node 3
    Node 4
    AAE exchange

    View Slide

  103. 103
    Active Anti Entropy
    Realtime
    Persistent
    Rebuilt Weekly

    View Slide

  104. Upcoming
    104

    View Slide

  105. 105
    Data Types (CRDTs)
    Strong Consistency (C/P)
    Solr Integration

    View Slide

  106. CRDTs
    106

    View Slide

  107. 107

    View Slide

  108. 108

    View Slide

  109. 109
    Counters
    Sets
    Maps

    View Slide

  110. 110
    Counters
    Sets
    Maps

    View Slide

  111. 111
    1 1 1

    View Slide

  112. 112
    1 1 1

    View Slide

  113. 113
    1 1 1
    2 4
    +1 +3

    View Slide

  114. 114
    1 1 1
    2 4
    +1 +3

    View Slide

  115. 115
    1 1 1
    2
    {2,4} {2,4} {2,4}
    4
    +1 +3

    View Slide

  116. 116
    (a,1) (a,1) (a,1)

    View Slide

  117. 117
    (a,1) (a,1) (a,1)

    View Slide

  118. 118
    (a,1)
    +1 +3
    (a,1) (a,1)
    (a,2) (a,1),(b,3)

    View Slide

  119. 119
    (a,1)
    +1 +3
    (a,1) (a,1)
    (a,2) (a,1),(b,3)

    View Slide

  120. 120
    (a,1)
    +1 +3
    (a,1) (a,1)
    (a,2) (a,1),(b,3)
    (a,2),(b,3) (a,2),(b,3) (a,1),(b,3)

    View Slide

  121. 121
    (a,1)
    +1 +3
    (a,1) (a,1)
    (a,2) (a,1),(b,3)
    (a,2),(b,3) (a,2),(b,3) (a,1),(b,3)
    =5

    View Slide

  122. 122
    Counters
    Sets
    Maps

    View Slide

  123. 123
    Counters
    Sets
    Maps

    View Slide

  124. Consistency
    124

    View Slide

  125. 125
    A/P or C/P
    Per Bucket Choice

    View Slide

  126. 126
    Single Key
    Optimistic CAS
    Counters/Sets/Maps

    View Slide

  127. 127
    Multi Paxos
    Disjoint Ensembles
    Scalable

    View Slide

  128. Node 2
    Node 1
    Node 3

    View Slide

  129. 129
    1 2
    3
    4
    5
    6
    7
    123

    View Slide

  130. 130
    1 2
    3
    4
    5
    6
    7
    123
    234

    View Slide

  131. 131
    1 2
    3
    4
    5
    6
    7
    123
    234
    345

    View Slide

  132. 132
    1 2
    3
    4
    5
    6
    7
    123
    234
    345
    456

    View Slide

  133. 133
    1 2
    3
    4
    5
    6
    7
    123
    234
    345
    456
    567...

    View Slide

  134. 134
    1 2
    3
    4
    5
    6
    7
    123
    Ensembles
    234
    345
    456
    567...

    View Slide

  135. 135
    Ensemble
    Indepedent Paxos Group
    Leader Election
    Serializability

    View Slide

  136. 136
    write 2
    1
    1
    1
    write 2
    leader

    View Slide

  137. 137
    write 2
    2
    2
    2
    write 2
    leader
    2
    X

    View Slide

  138. 138
    3
    3
    3
    write 3
    leader
    3

    View Slide

  139. 139
    3
    3
    3
    leader

    View Slide

  140. 140
    +1
    1
    1
    1
    +1
    leader

    View Slide

  141. 141
    2
    2
    2
    +1
    leader
    2

    View Slide

  142. 142
    3
    3
    3
    leader
    3

    View Slide

  143. 143
    3
    3
    3
    leader

    View Slide

  144. Riak + Solr
    144

    View Slide

  145. 145
    Next Generation Search
    Full Distributed Solr
    Embedded in Riak
    Active Anti Entropy

    View Slide

  146. End
    146

    View Slide

  147. 147
    We hiring!

    View Slide

  148. View Slide

  149. View Slide

  150. ricon.io

    View Slide

  151. Questions?
    151

    View Slide