L7: SLIs SLOs and SLAs
Вставка
- Опубліковано 15 жов 2024
- When we build reliable computer systems, we need a language to describe how reliable our system is, how reliable it is expected to be, and what we will do when it fails to meet expectations. This video explains the language distributed systems designer use to talk about reliability.
One correction: On the "Setting your SLO" slide it should read "SLO is greater than reaction time"[1]. It would be silly to have your SLO be shorter than the time it takes you to react. :-)
[1] Thanks to Mihailo Bjelic for spotting this bug! (Apparently I'm not allowed to put a greater than sign in my video description. UA-cam no longer lets you annotate a video directly. Oh well.)
See www.distributed... for more information about this series, the class project, and links to slides.
This has to be the best video on UA-cam about sla, slo, and sli. Thank you.
Glad you liked it!
Finally some understandable lectures on distributed systems, Thanks Chris
Glad it was helpful!
I wrote a beautiful bit of code in HTML last year. It prints “hello world” in the browser. Not a single bug. I’m so proud 😋💃🏽
Lol jokes aside, you’re awesome, and thank you! I’m working on a new blockchain algorithm and have problems wrapping my head around mathematical ways to achieve consensus among nodes. You showed us a naive solution to the Byzantine Generals Problem in another video, which helped, and I’m looking forward to your discussion of the Paxos and Raft algorithms. Best wishes, and again, thanks for making this playlist. It’s really getting me up to speed! 😊🙏🏽✨
It's really interesting listening to this, just a week after all google services had a downtime of 44 minutes in one day ... Really good video series, thanks!
Thank you!
Great videos Chris! You deserve a lot more views!
Is this me, or this guy looks and sounds like Dwight Schrute?
Great content btw, thanks a lot!
Great series so far...Thanks for these Chris
Glad you enjoyed it
Thanks Chris. These videos are helpful. Do you still plan to complete the series?
Great explanation Chris, thank you very much!
My pleasure!
Great videos, thanks for all of them! Just a small remark regarding this one - shouldn't it be "SLO > reaction time" instead of "SLO < reaction time" (you discus that at 14:38)?
Oh, I was just about to suggest putting an annotation on the video, didn't know they don't allow that anymore!? Anyway, it's really a minor mistake/typo, and you have described it correctly and clearly in the voiceover, so I guess barely anyone will notice it... Thanks for all the videos once again, your sense of humor and nice examples made things a lot easier to understand! :)
I'm looking at this cards thing, seems like a nice feature (especially because they're also being shown on mobile devices), but I guess they didn't predict all of the common use cases.. They should probably add a "Comment" card (which would serve the purpose in your case), besides currently available ones (Video, Poll, Link etc)..
Wow, this was amazing! Thanks
Thanks!
@@DistributedSystems helped me get my job at Microsoft as an SRE!! Just started a week ago. Thank you so much, this video definitely helped me in the interview
What are examples five nine, or four nine systems? Do they exist? what for are they needed? Thanks!
Healthcare, aeronautic and powerplant systems just to name a few where reliability is critical.
@@MaximilianBerkmann Thanks for reply. Could you point me to where to find papers (or list of papers) achieving such results?
@@nyimbo3360 You're welcome. I don't know if there are papers on this. There's perhaps some on www.researchgate.net/
An example of a four-nines system would be Google's search product. (They almost, but not quite, reach five nines in their SLI.)
I was once asked in a meeting what reliability my service could promise to users. I answered, totally seriously, "nine fives".
It took a minute before someone stopped the discussion, turned back to me, and said "wait, WHAT did you say a minute ago?!?"
Excellent video!
Glad you liked it!