Well, Cloudflare DNS is software and all software has the potential to fail I suppose. We try very hard to test and validate so that we don’t break things, and improve when things grow old and brittle (we just finished a migration of a key DNS component to new architecture because the old system struggled to keep up some days… I think there is a blog post on it).
Should everyone have a copy of their DNS zone file backed up somewhere no matter who the DNS provider they are using is? Absolutely… well maybe not my mom because the 3 people who read her blog can probably wait 2 days for her son to get around to fixing it. But you get my point.
We’ve tried to make our DNS system as redundant as possible and as simple as possible from a code perspective to reduce the risk anything could go wrong, but Mr. Murphy has a sick and twisted sense of humor.
If I were running a large organization (and I’ve been in that position before), part of my DR strategy would be testing my redundant/failover solutions on a regular basis just in case. In terms of critical infrastructure Cloudflare’s DNS would probably be pretty low on my list compared to other systems I’d consider more likely to fail, but DNS is a critical piece of infrastructure so a backup of the records and at least a written plan for what to do in case of a disaster is probably a reasonable line item.