Comment by afavour
6 hours ago
Admittedly when I wrote that I was thinking about the recent AWS outage. Anecdotally, I asked friends and family about their experience and they assumed the internet was down. Almost everything at my work runs on Google cloud so we were still running but we observed a notable dip in traffic during the outage all the same.
> it is still bad
No doubt. But there’s a calculation to make, is it bad enough to spend the extra money on mitigations, to hire extra devops folks to manage it all… and in the majority of end user facing cases the answer is no, it isn’t.
Where I've worked and we've been in the cloud I've always promoted just running in one AZ, I run my own things in one Hetzner DC (hel1). I've done hybrid cloud as well and in that case we only have one AZ for the on-premise stuff anyways (plus offsite backup)
That one time when an AZ goes down and your infra successfully fails over to the other two isn't worth it for a lot of my scale companies, ops consultants seem to be chasing high cloud spend to justify their own high cost. I also factor in that I live in Sweden where most infrastructure outages are exceptionally rare.
Ofc it depends on what kind of company you are and what you're providing.