nanog mailing list archives
Re: Multi-homing with multiple ASNs
From: Jimmy Hess <mysidia () gmail com>
Date: Sun, 23 Nov 2014 18:49:29 -0600
On Fri, Nov 21, 2014 at 8:49 AM, Curtis L. Parish <Curtis.Parish () mtsu edu> wrote:
I believe the state will modify their advertisements to add our ASN to the path but changes to advertising via the state network has to go through a design and change management process and then be scheduled into maintenance windows. Any attempts to balance the traffic via prepending will take weeks.
[snip] In other words, you are in effect not in control of the advertisement of your prefix, therefore you practically don't actually have an autonomous system, you have the number technically, but not the administrative division that is intended to exist. An appropriate amount of time to push out any change needed to an announcement should be no more than 1 business day, but less than 2 hours in an emergency, to add extra impending or pull an announcement. I would call a change management process that requires any longer unacceptable, or not reflecting the reality of the importance of well-maintained optimal properly functioning network connectivity. You have what seems to be something very fragile, and you have very low configuration agility, since you cannot change your announcements as needed out through the state as you need them to. A stateful firewall, has no correct place outside the border of a multihomed network; by definition, to have a stateful firewall, there must be a single point of failure (on the stateful firewall element) at least for each unique load-balancing tuple. So I would call (in this case), the origination of your prefix by multiple ASes a bad thing. The protocol allows this, but the other constraints related to the situation are serious impediments that make the solidity multihoming seem improper or potentially precarious, in terms of the true originating AS' ability to function as an AS and manage their network -- -JH
Current thread:
- Re: Multi-homing with multiple ASNs, (continued)
- Re: Multi-homing with multiple ASNs Mark Tinka (Nov 21)
- Re: Multi-homing with multiple ASNs William Waites (Nov 21)
- Re: Multi-homing with multiple ASNs joel jaeggli (Nov 23)
- RE: Multi-homing with multiple ASNs Curtis L. Parish (Nov 24)
- Re: Multi-homing with multiple ASNs Dave Crocker (Nov 24)
- Re: Multi-homing with multiple ASNs joel jaeggli (Nov 30)
- Re: Multi-homing with multiple ASNs Mark Tinka (Nov 21)
- Re: Multi-homing with multiple ASNs William Herrin (Nov 21)
- RE: Multi-homing with multiple ASNs Curtis L. Parish (Nov 21)
- Re: Multi-homing with multiple ASNs William Herrin (Nov 23)
- Re: Multi-homing with multiple ASNs Jason Bothe (Nov 23)
- Re: Multi-homing with multiple ASNs Jimmy Hess (Nov 23)
- RE: Multi-homing with multiple ASNs Curtis L. Parish (Nov 21)