Justifying The Use Of Large Caches - Was: Re: anyone know what powers "google public DNS?"
Sabahattin Gucukoglu <[email protected]>
| Newsgroups | gmane.network.djbdns |
|---|---|
| Message-ID | <[email protected]> |
I'm thinking of setting up a dedicated machine for a mailer. The mailer has built-in DNS routines, which support doing entirely recursive queries. With such routines, it would be hard to justify the use of any caching name server or standard resolver routine. But ... if these 'ere caching name servers are widely-enough used, suddenly the use of such resolvers and name servers seems relevant. I have an OpenDNS subscription to help my mother avoid bad things. I can add the IP address of the mailing machine. Then, I can tell the mailer to use the address of the caching resolver, or better yet, the local host where dnscache is running, and dnscache (which OpenDNS recommends, BTW) will be forwarding to the popular name servers. My question is: what shall I do? Ultimately, it seems we have to decide whether having local caching name servers will save us time, and we have to depend mostly in that regard on whether or not lookups need to be done recursively by the upstream server. If they do, we're wasting time; if they don't, we're saving it. So what kind of a bet is it that Google, or OpenDNS, will attract that kind of usage? I really would love to make use of cached results if possible from local servers (OpenDNS is here in London) with answers. Cheers, Sabahattin