Google DNS Resolver

[Trooper110]

http://googlecode.blogspot.com/2009/...s-new-dns.html

[DougBob]

Ok, somebody give me some pros and cons here, because my first reaction is BS.

[Trooper110]

Depends on how well their service works. I'd guess that Google probably has a pretty good cache of what's out their already. And I'd use it over, say, Comcast or another ISP's DNS servers. I already use Open DNS though and I'm happy with it, I don't feel the need to use Google DNS services.

[mapes]

I don't think it's BS. They say they cache continuously major domain names. Personally I love this

Validity: Google Public DNS complies with the DNS standards and gives the user the exact response his or her computer expects without performing any blocking, filtering, or redirection that may hamper a user's browsing experience.

No more open DNS $!#% redirects mangling my url.

So right now they index web pages in a search-able database. I'm assuming what they're doing is indexing DNS in an optimized version of they're web database

[DougBob]

So you're going to tell me this service is going to be so much faster than the regular DNS setup? What is it going to save, 500ms. I guess their setup is going to be something else we should put in their hands and rely on. Nope. DNS lookup is too important to me to desire a fraction of a second speed increase over reliability.
ISP DNS servers already have major domain names already cached from the root servers, so I'm not really buying off on the speed boost issue being worth it.

[Trooper110]

A further look: Is Google Public DNS safe? Look at the source ports - InternetNews:The Blog - Sean Michael Kerner

[mapes]

ISP DNS servers already have major domain names already cached from the root servers, so I'm not really buying off on the speed boost issue being worth it.

There's a TTL on that cache entry and it times out and is removed from the DNS. Which brings up a secondary point how long do they store the entry and what happens if it changes. Maybe the hash the entry and do periodic lookups to see if it changes.

[Trooper110]

Benchmarking:
Google DNS Benchmarking and Rationale

[DougBob]

Yes you are correct mapes, there is a TTL on the cache and it's there for a reason.

Current DNS functions perfectly fine!
This whole thing from google is bogus. There is no need to for their public DNS. The DNS structure in place functions sufficiently and as designed.
I have been working with DNS for a long time and can tell you for 100% certainty this whole speed thing is is not true.
The performance enhancement that needs to take place is the U.S. internet backbone it's self.
We haven't got to a point where the default 2 second DNS lookup timeout is causing a problem.
How many of you have ever checked you nslookups and seen a high percentage of nslookup timeouts due to the DNS server reponse, none of you for sure.
When and if you see the timeout it is due to network hickups not DNS.
After looking at the article and the facts I still say BS on this google issue.
The only valid use for the google proposal is firewalled intranets that use DNS to block resolution of certain domains and this would allow a person with the appropriate permissions to change their TCP settings to use the google DNS instead of the intranet.
This is just another fine example of someone exploiting the lack of knowledge of others for their own gain. I don't know what their motives are but I don't trust them and will never allow something like this to be used within any domain I have an influence over. From a security stand point, I'll block it as a security risk.

[mapes]

Hmm it seems to me a far jump from they're DNS not being faster to them being nefarious.