Distributed Systems 5.3: State machine replication

Sdílet
Vložit
  • čas přidán 27. 07. 2024
  • Accompanying lecture notes: www.cl.cam.ac.uk/teaching/212...
    Full lecture series: • Distributed Systems le...
    This video is part of an 8-lecture series on distributed systems, given as part of the undergraduate computer science course at the University of Cambridge. It is preceded by an 8-lecture course on concurrent systems for which videos are not publicly available, but slides can be found on the course web page: www.cl.cam.ac.uk/teaching/212...

Komentáře • 14

  • @levimk101
    @levimk101 Před 3 lety +11

    Thank you so much for creating this series. I've been reading Building Data-Intensive Applications and it is a wonderful book. Keep up the good work!

  • @zhou7yuan
    @zhou7yuan Před 2 lety +4

    State machine replication
    SMR [0:47]
    State machine replication (algorithm) [2:30]
    Closely related ideas [2:46]
    Limitations [4:16]
    Database leader replica [5:01]
    Replication using causal (and weaker) broadcast [7:02]
    commutative [7:26]
    total order | deterministic(SMR)
    causal | deterministic concurrent updates commute
    reliable | deterministic, all updates commute

  • @Joseph-co7uh
    @Joseph-co7uh Před 2 lety

    Beautifully explained. Thank you very much!

  • @nikolay6700
    @nikolay6700 Před 3 lety +4

    Эх. Надо еще раз смотреть.

  • @alihammadshah
    @alihammadshah Před 2 lety +1

    Love state machines 💗

  • @AnkitGarg
    @AnkitGarg Před 2 lety +2

    It is criminal how less views these videos have.

    • @Joseph-co7uh
      @Joseph-co7uh Před 2 lety

      Exactly. This lecture series is simply fantastic.

  • @tianwenchu1663
    @tianwenchu1663 Před rokem

    will replicas using FIFO-total order broadcast actually makes concurrent transactions (with writes) sequentially/serially executed on each replica with same order. So basically no need any locking on each replica?
    However, if each transaction including multiple reads and writes, can we still make these sequenced transactions to be parallel, at least for some reads?

  • @FarhadJabiyev
    @FarhadJabiyev Před měsícem

    I don't understand why at 09:41 you mentioned idempotencenl only for best-effort algo? All other algorithms have reliable network system model which can only be achieved with retry pattern. And if retry isnin place, then idempotence also should be, no?

  • @samlaf92
    @samlaf92 Před rokem

    Is FIFO a typo? It doesn't seem necessary since we already have a total order, and at @8:30 you start putting it in brackets ((FIFO)-total order broadcast). For eg., in blockchain-type SMR we clearly don't have FIFO since transactions can be reordered at will by the leader before being included in a block (eg. for extracting MEV).

  • @abcdef-fo1tf
    @abcdef-fo1tf Před rokem

    7:00 if we have two transactions being processed at the same time, does that mean they can interfere with each other?

  • @statcc0
    @statcc0 Před 3 lety

    03:59 Could you please recommend some introductory material into Blockchains, distributed ledgers, smart contracts? A set of lectures similar to yours would be ideal. I am curious to understand the basic principles of these protocols.

    • @daniilorain
      @daniilorain Před 3 lety +7

      1. Bitcoin and Cryptocurrency Technologies by Princeton University
      2. Introduction to Blockchain Technologies by INSEAD

    • @statcc0
      @statcc0 Před 3 lety +1

      @@daniilorain , thanks!