Re: [Isis-wg] Re: IS-IS as PE/CE routing protocol in BGP/MPLS/VPN

On Friday 16 May 2003 08:45, Manav Bhatia wrote:
> Stefano,
> 
> AFAIK a sham link is only required when two VPN sites that belong to the
> same area share a direct backdoor link. If no backdoor link exists between
> the sites then no sham link is required.
> 
> So what happens if i dont have that direct (backdoor) link between two
> different VPN sites or if there is no IS-IS adjacency over that backdoor
> link?

Then you do the usual propagation from/to isis/mp-bgp 
and I believe BGP has already all you need to make your 
own selection decision (localPref, MED).

Anyway, if you advertise the same prefix from different 
sites, I suppose is because these sites have some sort 
of common connectivity to that prefix... and what is 
the operational scenario where the same prefix is 
advertised with different metric-type ? If it is just a 
matter of preference you can just play with the metric 
value (with wide metric you have all you need)

s.

> 
> Regards,
> Manav
> 
> ----- Original Message ----- 
> From: "stefano previdi" <[email protected]>
> To: "Manav Bhatia" <[email protected]>; "lidefeng" <[email protected]>
> Cc: <[email protected]>; <[email protected]>
> Sent: Friday, May 16, 2003 3:15 AM
> Subject: Re: [Isis-wg] Re: IS-IS as PE/CE routing protocol in BGP/MPLS/VPN
> 
> If I understand correctly the concept of the sham-link, the
> PE will receive a mp-bgp route containing the other end of
> the sham-link and the extComm describing it. This route need
> to be the best among other (possible) mp-bgp routes (with
> same RD) for the same prefix. If such route is best, then you
> won't really import it into the vrf but rather you will "use
> it" in order to add a (fake) adjacency into the isis lsp the
> PE will originate. Then you run spf as part of the isis
> instance running in the vrf and compute your routes. After
> spf computation you'll find that the prefixes of the remote
> site are reachable through the sham link (according to the
> metric of your links of course). So I don't believe you
> really need to change the bgp selection process.
> 
> Btw, I don't see any reference to TLVs 22 and 135 in the
> draft...
> 
> s.
> 
>