If the price is good, having a fallback or a few is much safer than reliance on a single dc. So much so that it matters less if one of dcs burns or whatever. Cloudflare issues are theirs - they can’t blame the dc.
1) not informing CF of unplanned maintenance on electrical
2) not informing CF that they had failed over to generator due to a ground fault on their PGE transformer
3) not informing CF that ALL their generators failed and were on backup batteries
4) not informing CF that power had completely failed within the DC
5) notifying CF that anything at all had happened for 48 minutes (which was after all the UPS batteries had died)
Yes, this is a one sided analysis. I haven't seen the Flexential one (is there one?).
https://blog.cloudflare.com/post-mortem-on-cloudflare-contro...
However, CF massively failed at the whole "entire DC goes down" mitigation thing. This time it was Flexential screwing up power, but it could've also been a fire, a flood, a tornado, an earthquake...
Once you get to CFs scale, you simply cannot hide behind your vendors anymore. CF has gotten big enough that we can fully expect them to just deal with it.