Post

TL;DR: Don't keep all your eggs in one basket! What pisses me off about this whole Google Cloud London is down issue is that its not supposed to be like this. We all know the cloud is just someone else’s computer/infra, that's fine I'm not shitting on "the cloud", just how companies use it.

The great part of the cloud is that its SOMEONE ELSE'S computer / infra. If you build your infra correctly then someone else's computer(s) going down ain't your all hands alarm to deal with. "Oh eur-west2 is down? Well eur-west4 is fine" Or "Oh the whole of Google Cloud is down. Well AWS is fine".

The whole point of TCPIP was to route around issues in the network. Back when we thought that telco central offices / exchanges were a single point of failure (they still are - but better then they were) we created solutions to the issue. But many have gone back to that SPOF mindset.

But we also don't min/max every other part of the company simply because of cost. We don't say "Because the sales teams phone lines aren’t at 100% utilisation Alice and Bob are gonna have to start sharing a mobile phone!

When IT comes knocking for a new server, accounts shouldn't be thinking "Do we have the budget for this server", but should be thinking "Do we have the budget if the existing server does down?"