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


To: Paul Vixie <vixie@vix.com>
Cc: dnsop@cafax.se
From: Brad Knowles <brad.knowles@skynet.be>
Date: Mon, 28 Apr 2003 23:23:38 +0200
In-Reply-To: <g3d6j7w4ci.fsf@sa.vix.com>
Sender: owner-dnsop@cafax.se
Subject: Re: draft-ietf-dnsop-serverid-01.txt

At 5:58 AM +0000 2003/04/28, Paul Vixie wrote:

>  in other words, before we can decide how to encode or solicit or carry
>  nameserver identity, we have to decide the conceptual meaning of identity
>  as applied to nameservers.  if i loadbalance by running two processes on
>  a dual-processor system, but they respond to the same address/port combos
>  even though each one could have its own configuration, how many identities
>  do i have?  that sort of thing.

	Indeed, when I set up the caching nameserver farm at AOL in '96, 
I had four DEC Alpha 4100s with four processors each, 4GB of RAM, and 
four copies of BIND 8 running on each machine, each process listening 
to a different virtual interface/IP address that was bound to the 
same physical address.  I benchmarked the processes, and each one 
could handle about 2000 queries per second, regardless of whether I 
was running one or four processes, or anywhere in-between.

	Dunno what happened to that farm.  I'll have to ask some friends 
& former co-workers who might still be there.


	I'm thinking we might be able to do this with hashes of public 
crypto keys.  Each process would need a unique public key/private key 
pair for successful secure control via remote processes (e.g., rndc 
and such like), and this key could be kept in a separate file for 
each instance of each nameserver.  You could either include the 
public part (or hash thereof) in each EDNS response, or include a 
portion.


	However, I fear that we're getting well away from operational 
issues, and this aspect of the discussion should probably be taken up 
within DNSEXT.

-- 
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 a message to <dnsop-request@cafax.se>.

Home | Date list | Subject list