I am noticing a fairly mundane trend in that the GitHub outages tend to coincide with periods of (likely) high demand. This feels to me like a capacity issue more than a software bug crisis. At 4am on Saturdays GitHub never has any problems.
Centralization combined with massive AI demand is almost certainly the problem. If you move to GitHub Enterprise (on your own servers), you can probably exceed the availability of the current public cloud environment.
It is surprising that GHES rarely comes up as an option. It is a really good answer to this particular issue and eliminates the need to retrain the team on new tools. I understand the inclination to get salty about a vendor and cast the proverbial baby out with the bath water, but we must resist our impulses if we wish to make decisions that the business will find agreeable. The incremental additional licensing cost of moving from public cloud to on-prem is roughly $15-20 per user seat. If you only have 10-20 people using GH today and are deeply struggling with the consequences of these outages, this seems like a pretty easy decision to make.
Centralization combined with massive AI demand is almost certainly the problem. If you move to GitHub Enterprise (on your own servers), you can probably exceed the availability of the current public cloud environment.
It is surprising that GHES rarely comes up as an option. It is a really good answer to this particular issue and eliminates the need to retrain the team on new tools. I understand the inclination to get salty about a vendor and cast the proverbial baby out with the bath water, but we must resist our impulses if we wish to make decisions that the business will find agreeable. The incremental additional licensing cost of moving from public cloud to on-prem is roughly $15-20 per user seat. If you only have 10-20 people using GH today and are deeply struggling with the consequences of these outages, this seems like a pretty easy decision to make.