OSPF Sham-link

[romccallum]

Does anyone know exactly how the ospf sham-link operates. By default bgp learned routes do not get a label assigned (only the next hop). OSPF sham-link host interfaces MUST be advertised by BGP and not the ospf process. This says to me that the sham-link CANNOT be lable switched throughout the core network. However every document not very clearly always states the sham-link will follow the label forwarding table.

snippit below

The source and destination IP addresses

must belong to the VRF and be advertised by Border Gateway Protocol (BGP) to remote PE routers. The

sham-link endpoint addresses should not be advertised by OSPF.

Anyone any decent explanation of why this is.

[gmarogi]

The sham link is used to perfer the path via MPLS BGP VPN backbone instead of the backdoor link when it exists between two sites. The sham link is just like a virtual link and are configured between the 2 PE devices that connect to CE devices which have a backdoor link between them. Loopback interfaces are used to configure the sham link on the PE devices, and these are advertised thru BGP so that the PE devices can reach these addresses only thru the backbone and not included in OSPF so that the PE devices will not try to reach them through the backdoor link. Since the addresses are learned via BGP and there is a label assoicated with the BGP next-hop for these addresses (which might be another loopback or a physical address on the PE devices), the sham link between the PE devices can be label switched and OSPF routes are exchanged over this sham link ro prefer the routes via the backbone.