Laura   

@nimbinatus

Dev Advocate 🥑 , Pythonista, DevOps practitioner. Recovering Earth and atmo sci educator, cloud aficionado. My words are mine.

Austin, TX
Vrijeme pridruživanja: siječanj 2012.

Tweetovi

Blokirali ste korisnika/cu @nimbinatus

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @nimbinatus

  1. proslijedio/la je Tweet
    prije 15 sati

    Registration for TX primaries ends in hours! Still time! I'm holding down the station at Westlake . Don't miss your chance to use your voice!

    Poništi
  2. proslijedio/la je Tweet
    prije 22 sata

    Can't wait to have our in the Bay Area this week to talk about the future of and at 🥳 Register now!

    Poništi
  3. proslijedio/la je Tweet

    Have something to say but not place to say it? Is it about DevOps? Then I have a deal for you! CFP is still open and we need Lightening Talks and Track Talks!

    Poništi
  4. proslijedio/la je Tweet
    31. sij

    The videos from our very first are up on YouTube! Thanks to all our speakers, they're an all-star cast, check 'em out...

    Poništi
  5. proslijedio/la je Tweet
    30. sij

    Feynman Lessons for Learning: 🧠 1) Understanding is more important than memorization! 2) Learn principles, not formulas. 3) Ask questions! 4) Read Books every day. 5) Teaching is a powerful tool to learning.

    Poništi
  6. P.S. - Have fun trying out formal methods for distributed concurrent systems. The bank transfer example immediately came to mind because I read ’s book, which was recommended by when introducing me to formal methods.

    Prikaži ovu nit
    Poništi
  7. I hope this brain dive gives you appreciation for dev teams building these systems, QA teams ensuring the systems work, infra/ops/SRE teams dealing with these systems every day, and security teams keeping everyone else out of these systems. And all-in-one teams? 😎 You win.

    Prikaži ovu nit
    Poništi
  8. No wonder both systems are difficult to understand at scale!

    Prikaži ovu nit
    Poništi
  9. So Earth’s systems are chaotic because we have many variables providing input to determine each state of the system--each snapshot in time. Distributed concurrent systems have as many variables, and we examine both systems in a similar way: States and chaos theory.

    Prikaži ovu nit
    Poništi
  10. However, we can model the concurrency with formal methods like TLA+, identifying spots where we have deadlock: A special term meaning the model can no longer validate that a specific snapshot in time, or state, holds true for all assumptions.

    Prikaži ovu nit
    Poništi
  11. In distributed bank transfers, many servers send data simultaneously as people send money from one place to another. Redundancy in the datasets and two-phase commits are some tools to ensure the data is valid even if servers crash. All the data flowing simultaneously is chaotic.

    Prikaži ovu nit
    Poništi
  12. Want to really mess with your brain? Let’s make these systems we just examined not just distributed, but concurrent, meaning they can execute steps and processes in parallel or out of order and still end up in the same place. A classic example? Bank transfers across systems.

    Prikaži ovu nit
    Poništi
  13. However, despite all of these variables, we can still make decent predictions about how these systems act. We’re able to predict and account for errors and failures at specific points through modeling and engineering.

    Prikaži ovu nit
    Poništi
  14. Humans are unpredictable at some level. Here, we’re dealing with multiple levels from the dev that originally wrote the code to the infra person working the system. Add in a hacker trying to get in, and the randomization of the data in the system is even more impressive.

    Prikaži ovu nit
    Poništi
  15. When you think about it, the amount of data flowing in this system is astounding! And it’s chaotic because of all of the inputs--the variables--coming in from all sides. But what about the human part of the equation?

    Prikaži ovu nit
    Poništi
  16. What about the network? Is the backbone heavily trafficked so the data is caught at a choke point like water at a dam? What’s the network topology like? You can run up and down the OSI model and find more and more real-world variables:

    Prikaži ovu nit
    Poništi
  17. But this isn’t a theoretical system! Just like in Earth’s systems, you’re dealing with many variables. Let’s examine the physical hardware on the first server: Is the processor heating up? How good are the soldered connections on the motherboard? Are any of the cables damaged?

    Prikaži ovu nit
    Poništi
  18. Let’s say there are two servers wanting to talk between datacenters hundreds of miles apart. The first server sends a packet to initiate a handshake. In a theoretical system, that packet quickly and completely appears on the second server, which then follows protocol.

    Prikaži ovu nit
    Poništi
  19. Distributed systems are similarly chaotic. How so? Well, just because computers are involved doesn’t mean the systems are without chaos. After all, humans are interacting with them! Any computer system is chaotic, really, but distributed systems are particularly affected.

    Prikaži ovu nit
    Poništi
  20. We can make predictions, to a point, based on input from most of the variables and our innate ability to see patterns--in this case, we use math 😉

    Prikaži ovu nit
    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·