[j-nsp] LSP types in M-Series

[Rafal Szarecki (WA/EPO)]

Piotrek's  example loks well.

So You have per-class multiple transport-LSP between PE, but single VPN-LSP (inner label) spanned over multiple transport-LSP.
Of course use of multiple t-LSPs introduce probability of frames re-order, but this should be not a probem, because according to DiffServ, sequence of packets has to be mantained only in PHB Sceduling Class (EF, AF1.X, AF2.x...).

Rafal Szarecki JNCIE

skype me <callto://Rafal_Szarecki/> 

  

> -----Original Message-----
> From: juniper-nsp-bounces at puck.nether.net 
> [mailto:juniper-nsp-bounces at puck.nether.net]On Behalf Of Pedro Fortuna
> Sent: Friday, October 28, 2005 6:10 PM
> To: Piotr Marecki
> Cc: juniper-nsp at puck.nether.net
> Subject: Re: [j-nsp] LSP types in M-Series
> 
> 
> Thanks Piotr,
> It does not seem too hard. Its much easier to grasp when you 
> have an example :)
> Best Regards,
> Pedro Fortuna
> 
> On 28/10/05, Piotr Marecki <peter at mareccy.org> wrote:
> >
> > >> You can classify packets based on 802.1p and forward 
> them to different
> > >> LSP
> > >> with class based forwarding
> > >> ( doesn't have to be DS-TE , works also with normal TE) 
> - works great on
> > >> E-FPC. But it is not different l2circuit - y
> > >> ou can choose LSP but not l2circuit since logical 
> interface can be
> > >> assigned
> > >> to only one l2circuit
> > >
> > > Can you please provide an example, or a reference where I 
> can see an
> > > example of this type of setup?
> > >
> > > Thanks,
> > > Pedro Fortuna
> > >
> >
> > Sure . Consider router PE1 being igress of p2p martini l2 
> circuit with
> > following configuration :
> >
> > bubu at PE1 # show protocols l2circuit
> > neighbor 10.127.0.36 {
> >     interface fe-0/1/1.446 {
> >         virtual-circuit-id 446;
> >         community l2c1;
> >     }
> > }
> >
> > Communtiy l2c1 is used for further reference within policy 
> statement when
> > exporting to forwarding table
> > bubu at PE1 # show policy-options community l2c1
> > members 65001:1;
> >
> > Now , lets configure two different LSP for 10.127.0.36 
> egress , one used for
> > best effort BE , second for expeditied forwarding
> > EF - actual lsp topology is irrelevant , however LSP names 
> will be used in
> > cos-map to split traffic :
> >
> > label-switched-path to-gw4-ef {
> >     to 10.127.0.36;
> >     primary gw4-ef;
> > }
> > label-switched-path to-gw4-be {
> >     to 10.127.0.36;
> > }
> >
> > Now , assuming following COS config :
> > classifiers {
> >     ieee-802.1 ieee802 {
> >         forwarding-class be {
> >             loss-priority low code-points 001;
> >             loss-priority high code-points 000;
> >         }
> >             forwarding-class ef {
> >             loss-priority high code-points 100;
> >             loss-priority low code-points 101;
> >         }
> >       }
> >     }
> > forwarding-classes {
> >     queue 0 be;
> >     queue 1 af;
> >     queue 2 ef;
> >     queue 3 nc;
> > }
> > We need to associate classifier with l2circuit ingress interface :
> > bubu at PE1 # show class-of-service interfaces fe-0/1/1
> > unit 446 {
> >     classifiers {
> >         ieee-802.1 ieee802;
> >     }
> > }
> > and configure cos-map
> >
> > bubu at PE1 # show class-of-service forwarding-policy next-hop-map l2c
> > forwarding-class be {
> >     lsp-next-hop to-gw4-be;
> > }
> > forwarding-class ef {
> >     lsp-next-hop to-gw4-ef;
> > }
> >
> > Finally , we need policy-statement for forwarding table :
> >
> > bubu at PE1 # show policy-options policy-statement cbf-l2c
> > term t1 {
> >     from community l2c1;
> >     then cos-next-hop-map l2c;
> > }
> > bubu at PE1 # show routing-options forwarding-table
> > export cbf-l2c ;
> >
> > Notice however that this is one l2circuit with different LSP per
> > forwarding-class. Currently you can't associate
> > different l2circuit with every forwarding-class , since 
> every incoming
> > interface can only have one l2circuit binded to it
> > ( well , it's p2p , nothing more )
> >
> > best regards
> >
> > Piotr Marecki
> >
> >
> >
> >
> >
> >
> >
> 
> 
> --
> Cumprimentos,
> Pedro Fortuna
> 
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> http://puck.nether.net/mailman/listinfo/juniper-nsp
>