Earlier quoted context omitted.
The problem is bad design and/or integration of components. VMS clusters over leased lines went years without outages. Record was 17 years at a rail yard. They also did rolling upgrades across both different OS versions and CPU architectures . The methods are public if any competitors want to match or exceed them: https://support.hpe.com/hpsc/doc/public/display?docId=emr_na... These modern cloud services using Linux…
Modern apps are designed to deal with part of them being down - it is the core upgrade strategy. VMS clusters were (and modern equivalents are) awesome, but they have a different speed of innovation / development. It is a trade off - perfect uptime, but that new feature could take a year, or 99.9% uptime, but it can be written and deployed tomorrow.
https://www.youtube.com/watch?v=4fFDFbi3toc
Generally, on top of code reviews, one can get high return with minimal labor and hardware with a combo of Design-by-Contract, contract/spec/property-based testing, low-F.P. static analysis, and fuzzing with contracts in as runtime checks. That's my default recommendation.
In Github's case, they also might have access to both closed and open tools that Microsoft Research makes. MSR is awesome if you've never checked them out. Two examples applicable to system reliability and security:
https://www.microsoft.com/en-us/research/project/ironclad/
https://www.microsoft.com/en-us/research/publication/p-safe-...
Plus some of their other tools in various states of usability: