1) DNS resolution stops working - use a different provider. 2) Registrar interface stops working - wait till it starts working.
You were trying to solve problem 1 which required problem 2 to be solved first, but there's no way around that. This means that everyone has the same dependency issue with their registrar being a single point of failure. Hosting your actual DNS elsewhere doesn't change anything or prepare for this situation any better.
Using a registrar that is on separate infrastructure from your DNS host solves both problems.
So effectively, the registrar being up is the only thing that matters and there's nothing you can do about it.
You're betting on the fact that by using 2 services, they won't be down at the same time which might add some reliability, although most registrars aren't hosting their console sites on the same network as their DNS servers anyway so the odds of it happening are about the same. You were just caught in a bad situation.
True!
> most registrars aren't hosting their console sites on the same network as their DNS servers.
As a customer you have no ability to judge this. DNSimple was either hosting them on the same infrastructure or the DDoS was targeting both. I can't speak to what they do now, since this article is almost two years old.
By using one service for both components you're betting on the fact that they're doing the right things behind the scenes. IMO that is not a safe bet. For zero added monetary cost and effectively zero added management cost you can double your ability to weather that specific type of storm.
You can't change your nameservers if your registrar's management interface isn't resolving.