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


To: Robert Elz <kre@munnari.OZ.AU>
Cc: hardie@equinix.com, dnsop@cafax.se, aroot@ops.ietf.org
From: Mark.Andrews@nominum.com
Date: Fri, 03 Nov 2000 08:50:24 +1100
In-reply-to: Your message of "Thu, 02 Nov 2000 22:09:16 +0700." <560.973177756@brandenburg.cs.mu.OZ.AU>
Sender: owner-dnsop@cafax.se
Subject: Re: Anycast root metrics and analysis


	As the one who tuned BIND's RTT break points I agree with
	kre that it just needs to be better not best.  BIND's
	algorithm sets the first break point at 32 ms, then goes
	up in steps of 64 ms.  The primary purpose was to keep
	traffic within the continent, or closest continent in my
	case. Secondary to that it was to keep traffic in the local
	area of the continent.

	The real problem, which exists today, is to ensure that the
	infrastructure is capable to taking the traffic load without
	large queuing delays and without large packet losses.

	The root server operators can easily measure the losses at the
	server (if any).  It's much harder for them to see the losses
	in the network servicing the root servers.

	Mark
--
Mark Andrews, Nominum Inc.
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: Mark.Andrews@nominum.com

Home | Date list | Subject list