RAFT: A story on how clusters of computers keep your data in sync

Transcript

1. RAFT
   !1
   An adventure in keeping your data in sync
   Joshua Thijssen
   @JayTaph
   

   View Slide

2. What is X?
   2
   Set X = 5
   ok!
   5!
   

   View Slide

3. 3
   What is X?
   5! ...
   
   X ?!
   3.14
   

   View Slide

4. 4
   Gunter, we need a
   distributed consensus
   algorithm!
   Wenk!
   

   View Slide

5. How do we make sure that...
   !5
   ➡ Data stays consistent
   ➡ Servers can fail at any time
   ➡ Don’t rely on unreliable means (time, id’s,
   network etc)
   

   View Slide

6. Replicated State Machine
   !6
   #1 set Y=5
   #2 inc Y
   #3 set X = 3
   Journal / Log State machine
   Y=5
   Y=6
   Y=6, X=3
   

   View Slide

7. Where?
   ➡ etcd / CoreOS
   ➡ Zookeeper
   ➡ MongoDB
   ➡ ElasticSearch
   ➡ ...basically everything clustered
   !7
   

   View Slide

8. PAXOS
   !8
   Paxos is like, way
   too complex!
   

   View Slide

9. Diego Ongaro
   John Ousterhout
   !9
   In Search of an Understandable Consensus
   Algorithm (Extended Version)
   

   View Slide

10. RAFT
    !10
    

    View Slide

11. ➡ https://raft.github.io
    ➡ https://raft.github.io/raft.pdf
    !11
    

    View Slide

12. !12
    ➡ Proven to be correct
    ➡ Designed for simplicity
    ➡ Relatively easy to implement
    ➡ "Good enough" for most cases
    

    View Slide

13. ➡ Cluster of servers, usually 3 or 5.
    ➡ One single leader.
    ➡ Clients communicate with leader only.
    ➡ Leader sends logs to other members.
    !13
    

    View Slide

14. Three main pillars
    !14
    Leader
    Election
    Log
    Replication
    Safety
    

    View Slide

15. Leader Election
    !15
    "One leader to rule them all 

    (without the orcs and stuff)"
    

    View Slide

16. Follower
    Candidate
    Leader
    !16
    

    View Slide

17. !17
    I'm the
    leader!
    

    View Slide

18. !18
    ...
    Wenk!???
    

    View Slide

19. !19
    ...
    Vote for my
    as leader!
    

    View Slide

20. !20
    ...
    Wenk!
    Wenk!
    

    View Slide

21. !21
    ... I'm the
    leader!
    

    View Slide

22. !22
    Hey guys,
    I'm back!
    I'm the
    leader now!
    

    View Slide

23. !23
    I'm the leader
    for term 1!
    I'm the leader
    for term 2!
    

    View Slide

24. !24
    

    View Slide

25. View Slide

26. 26
    Servers Quorum Size Failure Tolerance
    1 1 0
    2 2 0
    3 2 1
    4 3 1
    5 3 2
    6 4 2
    7 4 3
    

    View Slide

27. Three main pillars
    !27
    Leader
    Election
    Log
    Replication
    Safety
    

    View Slide

28. Log Replication
    !28
    "How do we make sure that stuff also
    happens somewhere else"
    

    View Slide

29. !29
    1 2 3 4 5 6 7 8 9 10 11 12
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    3
    x←5
    3
    x←4
    index
    command
    term committed uncommitted
    

    View Slide

30. !30
    Leader
    1 2 3 4 5 6 7 8 9 10 11 12
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    3
    x←5
    3
    x←4
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    Followers
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    Leader receives commands and applies to log
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    

    View Slide

31. !31
    Leader
    1 2 3 4 5 6 7 8 9 10 11 12
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    3
    x←5
    3
    x←4
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    Followers
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    3
    x←5
    3
    x←4
    3
    x←5
    3
    x←4
    Leader pushes entries to the cluster
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    

    View Slide

32. !32
    Leader
    1 2 3 4 5 6 7 8 9 10 11 12
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    3
    x←5
    3
    x←4
    Followers
    Leader commits after majority and OKs to client
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    3
    x←5
    3
    x←4
    3
    x←5
    3
    x←4
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    

    View Slide

33. !33
    Leader
    1 2 3 4 5 6 7 8 9 10 11 12
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    3
    x←5
    3
    x←4
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    Followers
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    3
    x←5
    3
    x←4
    3
    x←5
    3
    x←4
    Followers receives logs and commitIndex from leader
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    3
    x←0
    3
    y←7
    3
    x←5
    3
    x←4
    3
    x←5
    3
    x←4
    

    View Slide

34. !34
    1 2 3 4 5 6 7 8 9 10 11 12
    1 1 1
    1 1 1
    1 1 1
    1 1 1
    1 1 1
    1 1 1
    1 1 1 2 2 2
    4 4 4 4
    4 4
    4 4
    4
    4 4
    4 4 5 5
    5 5
    5 5
    5 5
    6 6 6
    6 6
    6 6 6 6
    6 6 6 7 7
    3 3 3 3 3
    a
    b
    c
    d
    e
    f
    

    View Slide

35. 1007 1008 1009 1010 1011 1012
    24
    x←5
    24
    x←4
    24
    x←2
    24
    y←4
    1 2 3 4 1006
    1
    x←3
    1
    x←1
    1
    y←9
    2
    x←2
    23
    y←7
    ...
    Snapshots
    last included index: 1006
    last included term: 23
    Y=7, X=4,Z="aap"
    

    View Slide

36. Three main pillars
    !36
    Leader
    Election
    Log
    Replication
    Safety
    

    View Slide

37. safety
    !37
    

    View Slide

38. !38
    ➡ #1 For any given term, there can be at
    most one leader.
    ➡ #2 A server will vote for at most one
    candidate per term.
    ➡ #3 If two logs contain an entry with the
    same index and term, both logs are
    identical up to the given index.
    

    View Slide

39. !39
    ➡ #4 The leader in a term contains all the
    entries committed in previous terms.
    ➡ #5 Leaders always append new entries to
    logs. Entries are in leaders are never
    overwritten or deleted.
    ➡ #6 If server S1 applies committed log
    entry 'e' at index 'i' to its state machine,
    then no other server can apply a different
    committed log entry at index 'i' to theirs.
    

    View Slide

40. PHP?
    ➡ Actually, sort of
    ➡ https://github.com/Waqee/Raft-php
    ➡ https://github.com/JayTaph/raft
    !40
    

    View Slide

41. 41
    Questions?
    

    View Slide