[j-nsp] cspf to secondary loopback

[Rafal Szarecki (WA/EPO)]

Of course.

Lets me explain.

Whe you create 2 LSP (lsp-1 and lsp2) on R2 toward loopback address of R1, You will have two equaly good routers toward this R1-loopback in inet.3. Then BGP, when execute recursive lookup for BGP_NEXT_HOP for given prefix, has to choose one of two LSPs.

So, we hab=ve load-balancing scenario on the table.
By default this is per-perfix load balancing. You can madify this to "per-packet" what is effectivly per-flow load ballancing. But this is not what you expect.

The ony way to controll election of next-hop in case of 2 or more equaly good lsp is writting policy under [routing-option forwarding-table]

I understand that you have 2 clasese of prefices annouceb by iBGP from R1 to R2.What is differebtiator? Let assume existence of community 100:100 or 100:200

(I write from head so there can be errors)
[policy-options policy-stetement class1-class2]
term class_1 {
	from {
		protocol bgp;
		community class_1;
	}
	then {
		install-net-hop lsp lsp-1;
		accept;
	}
term class_2 {
	from {
		protocol bgp;
		community class_2;
	}
	then {
		install-net-hop lsp lsp2;
		accept;
	}
}
[policy-options]
community class_1member 100:10
community class_2 member 100:200

	

Please note that is lsp-1 or lsp2 will be down, then whole term is effectivly ignored, so default policy is executed.
Default policy = load ballancing.

In example I use community, but this can be anything - prefix-list, routefilter, ASpath, ....

You of course can write more terms to e.g. reject or whatever...


Rafa³ Szarecki JNCIE #136

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 Stefan Mink
> Sent: Monday, August 01, 2005 12:14 PM
> To: Piotr Marecki
> Cc: juniper-nsp at puck.nether.net
> Subject: Re: [j-nsp] cspf to secondary loopback
> 
> 
> Hi,
> 
> Piotr Marecki wrote:
> > If you dont't want to play with forwarding-table export 
> policies and still want to use CSPF and multiple BGP 
> NEXT_HOPs there is
> > alternative solution with install x.x.x.x/32 under 
> [protocols mpls label-switched-path xxx] stanza. Simply you can define
> > multiple addresses on lo0 which then you can use with your 
> multiple LSP's on the head-end with install command. Then your
> > BGP prefixes will be forwarded through relevant LSP , 
> depending on NEXT_HOP attribute. I know it's kinda 
> complicated solution , 
> > but at least you can use CSPF.
> 
> but since I set up the LSP with the primary loopback address,
> packets for prefixes with that nexthop would travel through
> the lsp as well which is exactly what I don't want to have.
> I'd like to be able to do separate TE for the individual NHs...
> 
>    tschuess
>              Stefan
> -- 
> Stefan Mink, Schlund+Partner AG (AS 8560)
> Primary key fingerprint: 389E 5DC9 751F A6EB B974  DC3F 7A1B 
> CF62 F0D4 D2BA
>