[27806] in North American Network Operators' Group
Re: Alternative to BGP-4 for multihoming?
daemon@ATHENA.MIT.EDU (Alex Bligh)
Tue Mar 14 11:59:08 2000
From: Alex Bligh <amb@gxn.net>
To: Randy Bush <randy@psg.com>
Cc: brandon@rd.bbc.co.uk (BrandonButterworth), nanog@merit.edu
In-reply-to: Your message of "Tue, 14 Mar 2000 08:36:42 PST."
<E12UuJG-00042d-00@rip.psg.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Tue, 14 Mar 2000 16:56:47 +0000
Message-Id: <E12Uuch-0004ii-00@sapphire.noc.gxn.net>
Errors-To: owner-nanog-outgoing@merit.edu
>
> > sure but which one are they most likely to be using for their
> > resolver (in firewalled environments it may be none of those)?
>
> the caching server provided by the dialup global isp.
Be fair to the guy - the caching server (not authoritative server)
is normally net.near the dialup client. It may be (mentioning
no global ISPs we know) that the call is hauled halfway across
the US / World to get to the one large datacenter containing
both the caching DNS server and the connectivity to between that
provider and the internet, but even so, serving pages from a machine
close to the data center is in that case better than serving pages
from a machine close to the user (unless you like tromboning data).
But yes, this technique sucks for regional content. And yes, ICMP
is not a good measure of connectivity.
--
Alex Bligh
VP Core Network, Concentric Network Corporation
(formerly GX Networks, Xara Networks)
