[68762] in North American Network Operators' Group
Re: 2001:590::/32 announced by both AS4436 (nLayer) and AS4474 (Global Village, no contact in whois, but seems to be nLayer...)
daemon@ATHENA.MIT.EDU (Joe Abley)
Tue Mar 16 12:59:09 2004
In-Reply-To: <200403161703.i2GH3Lt06691@karoshi.com>
Cc: NANOG (E-mail) <nanog@merit.edu>
From: Joe Abley <jabley@isc.org>
Date: Tue, 16 Mar 2004 12:58:12 -0500
To: bill <bmanning@karoshi.com>
Errors-To: owner-nanog-outgoing@merit.edu
On 16 Mar 2004, at 12:03, bill wrote:
> there is no problem with a prefix being
> announced by more than one ASN.
I am fairly sure that I have seen real-life issues with at least one
vendor's BGP implementation which led a valid route object with one
origin to be masked by another valid route object with a different
origin which was learnt earlier, a masking effect that continued even
after the original masking route was withdrawn.
I don't have any solid documentation or results of experiments to
support this, although it seemed very real at the time. It has always
led me to promote the conservative practice of advertising routes with
a consistent origin AS.
Bill: have you done any measurement exercises to determine whether this
is, in fact, an issue? Or was your comment above based on the protocol,
rather than deployed implementations of the protocol?
Joe
