Well the bad news is the H root servers were not available for over 18 hours. The good news is that practically nobody noticed. As it turns out a fiber cut and poor weather took out access to this cluster of root DNS servers. https://lists.dns-oarc.net/pipermail/dns-operations/2010-October/006142.html shows the explanation for the outage. While the outage had no direct impact on Internet users, it does point out the necessity of proper design for redundancy. Graph of the H availability: Cheers, |
Adrien de Beaupre 353 Posts ISC Handler Oct 3rd 2010 |
Thread locked Subscribe |
Oct 3rd 2010 1 decade ago |
This is awful. DNS is a robust protocol - but not a replacement for diversity (and backups).
Unless someone wrote "the service doesn't need to survive single-point failures." in the requirements document, it's woefully underengineered. |
DomMcIntyreDeVitto 45 Posts |
Quote |
Oct 3rd 2010 1 decade ago |
It's not underengineered. Nameservers use a hints file. Other roots use AnyCast.
http://en.wikipedia.org/wiki/Root_nameserver#Root_server_addresses |
DomMcIntyreDeVitto 2 Posts |
Quote |
Oct 3rd 2010 1 decade ago |
Sign Up for Free or Log In to start participating in the conversation!