While there are nominally 13 root nodes to the DNS system, they are anycast, meaning multiple hosts sharing the same IP address, geographically distributed. There are not literally 13 critical computers holding up the Internet, it's dozens in actuality. When a request hits a root node, usually the one physically closest and least loaded answers first. The root nodes have been DDOSed before, and so far have managed to weather the storm, because of the distributed nature of DNS.
The root servers themselves don't actually handle a lot of traffic. They largely delegate lookup of .com, .net, .org, etc, to other clusters of DNS servers which are in turn, redundant and anycast like themselves. And those .com, .net, .org, .gay, etc. servers in turn delegate to the various domain registrars for domains like facebook.com, kiwifarms.net, wikipedia.org, or nigger.gay. In some cases, the registrars themselves delegate lookups for a registered domain to servers owned by the holder of that domain.
That delegation why only AWS broke today instead of also Google and Microsoft. At Amazon's scale, it makes sense to own and operate their own DNS servers, instead of hoping GoDaddy can handle the scale of traffic involved in AWS having so many customers as to be a 1/3 of the entire fucking Internet. If those servers have issues, well lol, good luck doing anything on AWS.
And if 1/3 of the Internet is down because everyone runs their infrastructure on the part of AWS that broke today, well, better go find something to do that does not depend on the broken 3rd of the Internet.