Re: AS sets and sequences

Thank you Renan,

well I think it cannot be helped. However, there is still a question of why there are sets and sequences. When do sFlow exporters will export AS path data as a set (unordered) and when do they export these data as a sequence (ordered). There must be a reason for the existence of both type of sets, right? Actually, I would be ok with only sequences since they gave me the order of the next ASs to come.

Gregory

----- Original Message -----
From: Renan M Alves <renanmalves@gmail.com>
Date: Thursday, January 15, 2009 7:28 am
Subject: Re: [sFlow] AS sets and sequences

> Gregory,
>
> sorry but administrator can't force anything.
> sFlow will show what it see in the packet, without any treatment.
> May be you can use some tool to do that but I don't know one.
>
>
>
> On Wed, Jan 14, 2009 at 5:42 PM, <gregory@is.naist.jp> wrote:
>
> > Renan,
> >
> > thank you for your diligence.
> > Well, I do understand how the AS path is built but, sorry for the
> long> sentences I made in the previous mail, I just wanted to know
> if sFlow had a
> > say in whether to represent this AS path as a set (unordered) or
> as a
> > sequence (ordered). In the latter case, it may possible for the
> sFlow> administrator to force the representation of this AS path.
> > Isn't it?
> >
> > Gregory
> >
> > ----- Original Message -----
> > From: Renan M Alves <renanmalves@gmail.com>
> > Date: Thursday, January 15, 2009 4:29 am
> > Subject: Re: [sFlow] AS sets and sequences
> >
> > > Gragory,
> > >
> > > the order of ASs in a AS path is totaly determined by the routing
> > > protocol,commonly BGP, and represent the best path between two
> > > points (ASs).
> > > Sflow can't do anything about this, it just show what BGP decide.
> > >
> > >
> > >
> > >
> > > On Wed, Jan 14, 2009 at 3:29 PM, <gregory@is.naist.jp> wrote:
> > >
> > > > Hello,
> > > >
> > > > I may have missed something in the RFC, but I was not able to
> > > determine why
> > > > the destination AS path can sometimes be represented as a set
> > > (unordered set
> > > > of ASs) or sometimes as a sequence (ordered set of ASs). Is it
> > > bound to the
> > > > proximity of the destination? For example, in a case where the
> > > packet is
> > > > sampled far from its destination and there are still many ASs to
> > > cross, not
> > > > ordering the ASs in the destination AS path may save processing
> > > time (?).
> > > > Or is there a totally different reason to this distinction?
> > > > Moreover, is it possible to force the AS path to be either
> one or
> > > the other
> > > > in some sampling device implementations, especially routers? I
> > > can not
> > > > recall anywhere to configure the way the AS path is exported on
> > > the devices
> > > > I use. So I was wondering if it was automatically determined
> by an
> > > > algorithm.
> > > >
> > > > Well, my reasoning is way too long. Sorry.
> > > >
> > > > Thanks for the answer,
> > > >
> > > > Gregory.
Received on Wed Jan 14 17:21:20 2009