a) you had a pool of DNS server names, say 20, all with unrelated hostnames
b) you assigned 2 to each customer, randomly, when they configured a domain to use your servers.
Then, a DDOS attack would impact 10% of your customers instead of 100%. (Assuming other practices, like null routing the target until resolved)
.com was < 1 minute, .cm is > 55 minutes.
Their "update DNS server" page was acting a bit wonky, kept saying some of my nameservers were invalid when they weren't, but I eventually got them all switched.
This isn't a dig against Namecheap, it sounds like this attack is pretty bad, but for important domains Route53 just seems like a much better setup (geographically dispersed, different nameservers for each hosted zone, etc).
Good luck.
tamar (also at Namecheap)
That said, I've heard Cloudflare's DNS network is faster than many paid alternatives.
I'll probably continue with that setup, but just make sure my TTL is set fairly high.
If Namecheap is down, it does not get redirected.
My domain is on v1 now and its still working.
Instead, the second that this DDos hits is the second we have websites stopping working.
How is it that in this day and age we can't have distributed caches of DNS entries at our providers of full dns databases. I mean there can't be more than like a few billion dns entries in the world total, which fits easily in a modern desktop computers RAM.
If that is an underestimate, I can't believe a single modern server wouldn't be able to mirror the world's DNS queries for at least a providers worth of users.
I sort of hoped that a DNS client would just use an expired DNS result in case the servers would not respond, but perhaps that is naieve/dumb.
I assume you'd make sure the DNS records are the same in both DNS portals; and then add Route53 as 3rd & 4th nameservers with the first and second still being Namecheap?
If you want speed and readability I suggest switching to a paid DNS provider.
1: http://www.solvedns.com/dns-comparison/2014/01
BTW I'm not in any way affiliated, just like the service.
I am monitoring a few servers with DNS records. And the last week I have found all the servers unresponsive (by DNS, not tried directly) from time to time. And after an extensive amount of troubleshooting I am unable to find a problem.