[bartread]

This happens in software too.

Over the last 4 years our team has achieved a lot: huge numbers of valuable changes and improvements to our platform. But it's been much harder than it might otherwise have been because we've had to make those changes with the systems in use. Had we started from scratch, or been able to take downtime, there are a lot of projects we could have done much more quickly, but we had to keep the business running - it is, after all, what was and is paying all of our salaries.

[tomfanning]

Isn't this completely standard practice now?

[tialaramex]

24/7 is table stakes for many Internet companies, but lots of outfits which think of themselves as delivering that sort of service actually cheerfully carve out hours or even days of down time as "necessary".

One of my banks decided it was going to do a "major upgrade" one weekend. Advertised I think maybe 8 hours outage like hey, who needs a bank for eight hours right? And of course their team can't actually hit that schedule, but nobody wants to choose "Roll back, fall on my sword at breakfast time" so an hour after the end of that supposed 8 hour outage their telephone support were telling me it ought to be fixed "soon" and any problems are only "temporary" and I can try again in a few minutes.

They got it back later that day, no noticeable improvements and you can bet that even if there was some enquiry about what went wrong nobody learned anything from it. Like NASA after Challenger. And they will still send representatives to the IETF who will say well, we can't afford these random outages like you Internet people, we're a bank, we need high availability. And those representatives will look around wondering why everybody is laughing.

[bartread]

Not always. I've worked on all sorts with different companies.