Hi Adrian, 

> 
> >> Oh, could you also explain why it is not necessary to 
> advertise the 
> >> VPN ID.
> >> Or did I miss it?
> >
> > Certainly. BGP has all the necessary mechanisms to achieve both VPN 
> > membership distribution without a need to carry specific IDs that 
> > uniquely identify the VPNs.
> 
> Right. No specific ID needs to be carried, but it must be 
> possible for a PE to determine, for each VPN it serves, which 
> remote PEs (BGP peers) also serve that VPN, 
>and which 
> advertisements from those PEs apply to which VPN (i.e. into 
> which PIT to place the advertisements). Lastly, it must be 
> possible for the PE to determine which PIT to access when 
> setting up L1VPN connections.
>

Correct.
 
> > Since the type of l1vpn service we are interested in is a 
> port-based 
> > VPN service, from a control plane point of view all what is 
> needed is 
> > the ability for the remote PEs to discover the set of PEs that have 
> > VPNs in common and the set of ports associated with the VPNs.
> 
> Agreed.
> 
> > BGP allows that through the use of
> > BGP-MP and with the use of extended community (such as 
> route target) 
> > for the membership and topology distribution. This scheme 
> plays both 
> > the role of a VPN-ID and topology information distribution 
> without a 
> > need for an explicit advertisement of VPN-IDs.
> 
> OK. Understood.
> 
> Essentially, the "shared" configuration information across 
> the network is the route target, or rather the set of route 
> targets, that correspond to the L1VPN.
> 

Yes. Added with the mechanism of import/export route targets.

> > Do you (or why do you) think a VPN-ID is necessary in the discovery 
> > process?.
> 
> No I don't think it is necessary.
> It might have been helpful for diagnostic purposes. But since 
> it is possible to map between route target and L1VPN ID 
> within the management framework, it is not essential.
> 

Agreed here too. I can see for other functions within the VPN a 
VPN-ID maybe useful like management, etc.

I think we are in agreement on all of that. 

Thanks.
Hamid.


> Cheers,
> Adrian
> 
> 
> 
> 

_______________________________________________
L1vpn mailing list
L1vpn@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/l1vpn

Reply via email to