[31362] in North American Network Operators' Group
RE: Root zone change -- d.gtld-servers.net
daemon@ATHENA.MIT.EDU (Roeland M.J. Meyer)
Thu Sep 21 12:33:19 2000
Message-ID: <1148622BC878D411971F0060082B042C36FD@hawk.lvrmr.mhsc.com>
From: "Roeland M.J. Meyer" <rmeyer@MHSC.com>
To: Michael Shields <shields@msrl.com>, Austin Schutz <tex@off.org>
Cc: rdobbins@netmore.net, randy@psg.com, nanog@merit.edu
Date: Thu, 21 Sep 2000 09:28:41 -0700
MIME-Version: 1.0
Content-Type: text/plain
Errors-To: owner-nanog-outgoing@merit.edu
> From: Michael Shields [mailto:shields@msrl.com]
> Sent: Thursday, September 21, 2000 8:04 AM
> I think the best way to tell which TLD servers are most widely used
> would be to monitor the query traffic to the root nameservers; most
> operators will agree that the most actively trafficed domains are the
> most important to maintain reachability to. BIND's selection
> algorithm will bias this however, so you would need to take a sample
> across several or all.
This would be valid if, everyone actually used the root servers. Those
behind firewalls don't <plug>and neither do ORSC systems</plug>. They
have their own root servers. However, the largest hole will come from
the firewalled systems. Oh yeah, there are also those of us who
instantiated our own roots when NSI started hiccupping a few years ago.
Every NSI glitch has been transparent to us since then <grin>.
Because there is no single point of failure on the internet, there is
also no single point of monitoring. What you see is not all there is.
Also, the NSI notice did mean I had to update my copy of the rootzone,
since the TLD servers moved. Anyone running BIND, behind a firewall, had
to do the same.
