[boulos]

For us (Google), I think you're referring to: https://status.cloud.google.com/incident/compute/16017

which was caused by our own maintenance several weeks apart (the root cause description is really quite good).

I think the distinction people make implicitly is a 25 minute outage versus 8 hours. DDoS attacks suck, but they're just standard these days. As a customer though, any source of (network) outage usually has the same outcome: "my site is unreachable (and I don't care why)".

The reason we (and AWS and others) offer multiple datacenters/zones within a <1ms boundary (a compute "region") is so you can build a highly available app that can fail over with minimal degredation. For customers that were using App Engine Flexible environment with the regional spreading turned on, only some of their instances were affected, but their apps shouldn't have skipped a beat.

Linode is good at what they do, but any customer in Atlanta just had to wait this entire event out.

Disclosure: I work on Compute Engine.