[3749] in North American Network Operators' Group
Re: Customer AS
daemon@ATHENA.MIT.EDU (Henry Kilmer)
Mon Aug 19 15:41:12 1996
Date: Mon, 19 Aug 1996 15:27:31 -0400 (EDT)
From: Henry Kilmer <hank@rem.com>
To: Vadim Antonov <avg@quake.net>
Cc: hank@rem.com, jon@branch.com, hank@sprint.net, nanog@merit.edu,
randy@psg.com, smd@chops.icp.net
In-Reply-To: <199608191914.MAA02347@quest.quake.net>
Vadim Antonov writes:
>In my (rather extensive) practice, multihoming by itself is
>usually a major source of connectivity problems.
Agreed.
>Whoever arguing _for_ mulihoming for everyone forgets that
>taking more routing information in has dangers not present
>when you don't do routing yourself.
>
>I never saw any customer who had the ability to configure a
>multihomed site properly on their own; and most of the bogus
>routing information comes from multihomed customer sites.
>
>It is _much_ better to multihome to the same provider who then
>can take care of messy global routing.
Agreed.
The arguement here (if there is one) is that their is a demand in the
marketplace for multihoming to different providers and what is the
best way to treat these customers. Sprint's filtering is a good
arguement for having multiple Sprint connections or non at all. The
customer that is multihomed to Sprint and a different provider,
however, is still paying Sprint to move their bits around even if
their Sprint connection goes down.
I support all incentives to reduce the number of multihomed customers.
>--vadim
>
>PS A UPS for CPE usually fixes 95% of transmission problems.
> I've seen people willing to spend money on multihoming but doing
> things on commercial power.
And not plugging their routers into outlets on light switches would
also help.
-Hank
