Date: Mon, 13 Jan 2003 15:51:43 +0100 From: Brad Knowles <brad.knowles@skynet.be> To: Pete Ehlke <pde@rfc822.net> Cc: freebsd-chat@FreeBSD.org Subject: Re: Need advice on PHP and MySQL books Message-ID: <a05200f07ba4881233aa2@[10.0.1.2]> In-Reply-To: <20030113141542.GC2260@rfc822.net> References: <20030110234309.R12065@2-234-22-23.pyvrag.nggov.pbz> <3E1FF12B.5390D978@mindspring.com> <20030111144619.X22424@2-234-22-23.pyvrag.nggov.pbz> <a05200f22ba474b755a2a@[12.27.220.113]> <3E21FD22.38CD81BB@mindspring.com> <a05200f02ba47b68abed1@[10.0.1.2]> <20030113141542.GC2260@rfc822.net>
next in thread | previous in thread | raw e-mail | index | archive | help
At 8:15 AM -0600 2003/01/13, Pete Ehlke wrote: > Does anyone know the story about NSD? I've looked at it several times, > run it and played with it locally quite a bit, and found it extremely > interesting. But I've had a third-hand report that RIPE folks have said > (third hand, but this is the direct quote I got...) "the damn thing just > didn't work". Haven't been able to get more than that. I've benchmarked it, and it is *damn* bloody fast (see <http://www.shub-internet.org/brad/papers/dnscomparison/>). However, the way it works is to pre-calculate every possible query for the zone in question, and then to pre-generate every possible answer (coalescing as many questions and answers together as it can), and then to index all the answers from all the questions. They throw out all the "normal" authoritative name server tasks that would normally be done that are not strictly required for the kind of operations you would expect to see at a root name server. So, for a moderately large zone, it is incredibly fast, but can return somewhat strange answers (due to the way it optimizes queries and may ignore domain name compression or implement it differently than you would expect), and it certainly doesn't do anything like round-robin, etc.... But for much larger zones (like .nl, and especially the signed "test" version at .nl.nl), you just can't feed it enough memory. Or so I have been told, and by people at least one step closer to that process. Unfortunately, I learned some things about the way it works *after* the presentation I did at LISA 2002, so now my view of the program is not nearly so positive as it was then. Personally, I'd be much more interested to learn how the folks at Nominum managed to get CNS and ANS to be as incredibly fast as they are, and still manage to provide "full service" DNS. Sure beats the crap out of anything else I've ever seen. -- Brad Knowles, <brad.knowles@skynet.be> "They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety." -Benjamin Franklin, Historical Review of Pennsylvania. GCS/IT d+(-) s:+(++)>: a C++(+++)$ UMBSHI++++$ P+>++ L+ !E-(---) W+++(--) N+ !w--- O- M++ V PS++(+++) PE- Y+(++) PGP>+++ t+(+++) 5++(+++) X++(+++) R+(+++) tv+(+++) b+(++++) DI+(++++) D+(++) G+(++++) e++>++++ h--- r---(+++)* z(+++) To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-chat" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?a05200f07ba4881233aa2>