[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Nameserver worlds and Internet namespace

    Date: Thu, 26 May 88 20:43 EDT
    From: CJL@REAGAN.AI.MIT.EDU (Chris Lindblad)

	Date: Thu 26 May 88 12:05:06-PDT
	From:     TYSON@Warbucks.AI.SRI.COM (Mabry Tyson)

The marked similarity between your code and mine isn't accidental, BTW: I
started with yours, but wanted to deal with secondary servers. (credit where
credit is due).

    Actually, I have come up with another solution to the slowness of loading the
    internet namespace:  I got rid of it.  Since the mailer insists on using the
    domain database anyway, the internet namespace wasn't used much around here.
    Eliminating the Internet Namespace has improved the performance of our
    namespace servers significantly.  

    Now only if the resolver worked better...

This may be true, but we've found the resolver to be so bad, and UNIX
namespace resolution in particular to be bad (it never tells you what the
services are) that we even had problems resolving names at our local site
from the UNIX name server that tells the world about our site, let alone
remote stuff. I've seen relay.cs.net end up with different domain entries on
different machines at our site: some with a complete entry, though all the
services, in say, triplicate, and others with no services entered at all
(causing the mailer to crash).

Setting neti:*allow-dotted-host-names-in-namespace* to t seems to at least
allow things to get resolved in the internet namespace, otherwise, you can't
have such things there, and they won't get resolved there (as soon as the
host-parser sees a dot it just assumes it should ask the domain resolver, if
this parameter is bound to t it will at least check the namespace system).

Yes, the namespace server is slower, and the world is much bigger, and you
have to make sure the namespace is periodically updated, but finding an
entry in the INTERNET namespace is a lot faster than trying to get it
resolved, so thruput on the non-server machines is better, I've found, and
seems to be a virtual requirement for putting non-local host names (as, say,
aliases) in your mailboxes.text file.

Brad Miller		U. Rochester Comp Sci Dept.
miller@cs.rochester.edu {...allegra!rochester!miller}