3 ms·
If by “remove a few hubs” you mean level a few major colo’s then ok, but as far as I can tell there is not any single strand of glass or single switch or single
by ip26 5y ago
If by “remove a few hubs” you mean level a few major colo’s then ok, but as far as I can tell there is not any single strand of glass or single switch or single server that can take everything down with it.
- toast0 5y agoThe DNS root is pretty well distributed. There's 13 different IPs run by several different organizations, and AFAIK they're all running anycast these days. And anyway, the root zone is tiny, changes infrequently and could really be AXFRed a couple times a month and you wouldn't miss much. The larger tlds aren't quite as diversely hosted and certainly aren't amenable to long term caching, but it should take a major f up to break those too. Some of the minor tlds, even the more popular ones do screw up from time to time though.
- throw0101a 5y ago> As of 11/26/2021 10:53 p.m., the root server system consists of 1477 instances operated by the 12 independent root server operators. * https://root-servers.org https://root-servers.org * https://en.wikipedia.org/wiki/Root_name_server https://en.wikipedia.org/wiki/Root_name_server
- dylan604 5y agoYou don't need to level the colos, you just need someone to make a typo in a router config that gets deployed live so the whole colo is unreachable. How many times have we seen an AWS/CloudFlare/otherLargeProvider have this happen to them?