HE.net BGP origin attribute rewriting

Joe Provo nanog-post at rsuc.gweep.net
Sat Jun 2 10:43:15 UTC 2012


Last post on this topic for me. You seem to wish to argue 
against the lessons of history and the reality of running
a network on the global Internet.

On Sat, Jun 02, 2012 at 09:27:36AM +0200, Daniel Suchy wrote:
> On 06/02/2012 02:53 AM, Joe Provo wrote:
> > Cost and performance were merely two reasons someone may wish to prevent
> > remote parties from using origin to influence outbound traffic from my 
> > network. 
> As I mentioned already, it will influence that by another way. And this
> costs *you* more money - you have to pay for router with larger TCAMs,
> more memory, faster CPUs... and yes, deaggregation is very simple task
> for originating network - much easier than playing with the origin flag,
> which is not understanded widely.

The two issues are orthogonal. Deaggregating sources have 
been cost-shifting [in a highly visible and easily examined
and often trivially-filtered] manner for ages. There is no 
data to support the premis that touching origin creates more
of this behavior and plenty to refute it. Deaggregation
preexists and was always a problem with which one had to 
deal as supposed "needed TE" by those too cheap to build a 
proper network sadly became more acceptable over time.

A midspan network deaggregating someone else's prefixes is 
broken and gets called out, generally by the originator if 
they have a clue.
 
> > I can state it is not imagination when I encountered networks
> > doing this in the past for prefixes they were sourcing. To be clear - 
> > these were prefixes being sourced by a neighbor who was providing 
> > different origin codes on different sessions. Either they were [to
> > Nick Hilliard's point] using different kit and unaware of the differnt
> > implementations or [as evidence bore out] purposefully shifting traffic
> > without arrangement on links that were worse for me and in violation 
> > of the agreement we entered into when peering.
> 
> More specific prefix in addition to aggregate one visible only over
> specific peers will do the job, too. And will do that job better... but
> for what cost (not only to you)...?

See above. 

> > There certainly were historical reasons for treating origin as sacrosanct.
> > Time has marched on and those reasons are now *historical*, hence the 
> > quite reasonable updat eto the RFC. You seem to fail to understand that 
> > MED comes after origin on the decision tree, and therefore someone can 
> > influence traffic carriage without agreement.
> 
> You seem to fail realize other (easier) ways to influence traffic
> carriage. Deaggregation with selective route announcement is quite
> common way, many networks do that.

See above. 

Cheers,

Joe

-- 
         RSUC / GweepNet / Spunk / FnB / Usenix / SAGE / NewNOG




More information about the NANOG mailing list