Losing two 9's while gaining 45x the traffic

追加: 2026-05-16

[00:00:00]When GitHub talks about why there outages happen, like it's not cuz they're incompetent or because they're bad, it's because their traffic went up like 45 x in 1 year. And that's just like hard to manage. One of the GitHub's problems is that it was built as a SaaS first, and so one customer's downtime ends up affecting basically every other customer unless you're on enterprise.

[00:00:19]And a lot of people that don't want to do their own stuff or whatever else or like there's a lot of people who are using the sort of SaaS version of GitHub. And also like some of his enterprise stuff is not as fully sorted out, blah blah blah blah. And so like we've developed stuff in a way where it's bring your own cloud, and so you're running your own copy of our stuff yourself. And so if we had scaling problems which we're first of all just trying to architect this so that we don't, but like if one customer breaks things, it doesn't break everything for every customer cuz each customer has their own deployments of things. And so, you know, that's just like a more natural way of scaling up is that like, you know, you don't have a 1% downtime for one customer means 100% downtime for all customers when you're doing sort of the default SaaS kind of like platform sort of sent to the world. And so we want to make it sure that like customers out like if customers have a downtime it does not affect other customers and stuff like that that like kind of matters that we're doing stuff that's more come big company focused than the sort of like uh social network that GitHub is doing.