>https://seekingalpha.com/news/3442879-facebook-says-outage-caused-server-change-now-resolved📁
>"Server config change."
>Q
SoftwareEngineerAnon here. Facebook's claim that the outage was a result of a "config change" is 110% bullshit. I say that because it is industry standard to go through a process known as "A-B testing" where you stand up two or more production environments, then update one, and transition the load over time. If any problems arise, then you pull back and let the old instances handle the load like normal. That's problem ONE, which a company like FB would obviously anticipate (how much downtime have they had historically when changes occurred?). Problem TWO… There's a concept of PERFORMANCE TESTING before you deploy to production! This again is industry standard and ANY company with $$$ on the line would ensure this practice is adhered to. Basically it involves standing up a non-public production-liek environment, then you simulate a massive load against that environment, usually at least what you would expect at extreme peak traffic times. Usually you'd want to simulate a load 2-3x greater than what you'd expect to future proof your technology.
Q knows what I'm talking about.