"Jepsen II: Linearizable Boogaloo" by Kyle Kingsbury

Поділитися
Вставка
  • Опубліковано 9 лют 2025
  • What is ACID anyway, and why is everyone so uptight about it? When can't we have eventual consistency, and what do we sacrifice in exchange for stronger models? I've spent the last year trying to wrap my head around consistency in distributed systems, and testing databases to see how those consistency models play out in practice. In this talk we'll explore linearizability--one of the strongest consistency models for a concurrent system--move from an academic definition to an intuitive understanding, and see the ways in which databases succeed--and fail--to live up to their consistency claims.
    by Kyle Kingsbury (@aphyr) - Factual
    Kyle is the author of Riemann, Timelike, Jepsen, Knossos, and a host of other open-source libraries. He works as a backend engineer at Factual.

КОМЕНТАРІ • 4

  • @raptor1737
    @raptor1737 10 років тому +3

    Great talk. Every distributed systems engineer needs to watch this. Thanks Kyle.

  • @minikomi
    @minikomi 10 років тому +1

    Great talk!

  • @DaveCS2
    @DaveCS2 10 років тому +2

    12:50 -- Why is the second scenario not safe (not good)? I don't see any lines "falling off cliffs".

  • @victornoagbodji
    @victornoagbodji 10 років тому +3

    great presenter : )