because if everyone did it you would bring down the root servers.
the caching mechanism is not there for show...
Re:Google DNS was down !
by Nick-Craver on Monday October 13, 2014 @08:51AM
(#8448484)
This is simply not true - it assumes that everyone's forwarding server for an unknown record is the root servers directly, which isn't the case.
If this were true, my life would be much easier because DNS propagation would be much more predictable than it is today. One bad server in the chain serving obnoxiously high and invalid TTLs can ruin your day. It's not very common percentage-wise, but it certainly happens every time we switch DNS over for Stack Overflow.
Re:Google DNS was down !
by mike-cardwell on Monday October 13, 2014 @08:48AM
(#8448472)
"because if everyone did it you would bring down the root servers."
What figures are you basing that on? The TTL for the NS records for '.' is 6 days. Or are you just guessing? I think it would be perfectly feasible for everyone to run a local recursive resolver. As long as the change was gradual rather than over night.
And if that reason is you want to take advantage of a third parties cache, you can configure Unbound to forward requests onto another resolver, but also for it to fallback to doing it's own resolution if that third party resolver times out.
Unbound is easy to set up, and runs on Linux, OSX and Windows.