Provided the explanation of the topology is correct and Area 0 remains contiguous after the DC to DR link failure your DC to DR traffic will stay within Area 0 without any need for any filters. If your topology is such that there is no resiliency on your Area 0 interconnect and after DC to DR link failure both DC and DR have no neighbors left in Area 0 then there is potential of DC to DR traffic transiting the branch Area. In this case if I were you I would look at adding resiliency to your backbone Area. Atif ... View more

Ahmed, If you can post your OSPF logical topology showing Area layout it will be easier for us to provide an accurate response. Depending on how the Area layout is what you want can be met automatically by OSPF. For example if the DC and DR routers are ABRs and the branch area is different than the DC/DR LAN area then DC to DR LAN traffic will not traverse the branch area. Atif Ahmed Chowdhury wrote: hi, i have two core router one is DC end and another is DR end. every branch office is connected to this two routers. DC and DR router also connected through a vendor. my requirement is when the connection between DC and DR failed then i dont want that DC local lan reach the DR local lan through branch routers which means that the branch acts as a transit path. in this case DC and DR local lan dont need any communication. i have a solution about distribute list but it should be set in every tunnel interfaces which is very much cumbersome for my scenario. if anyone have any suggestions then please share with me. i also attached a diagram below............ ... View more

Irfan, By definition an EoMPLS service in port-to-port mode should let you do whatever you want. However, there is another variation of the service which is relaying a specific VLAN across sites. In that case you cannot carry multiple VLANs over the wire unless you do QinQ tunneling on your equipment. That being said in your scenario a lot depends on the capabilities of the SHDSL modem as well as how the ME3400 has been configured. You should clearly ask the provider that you need a port to port EoMPLS that should allow you to carry multiple VLANs across the provider backbone. How the provider configures it is up to them but in my opinion they should be able to offer this service barring any hardware/software limitation on the SHDSL modem side. Atif ... View more

msobier123 wrote: Atif, However, I haven't seen service providers implementing this command on the Core, we didnt use it before and I have confered with different people m the are not using it either on the Core. and the Second point, The Core needs to advertise the PEs addresses and IGP Labels for reachability and to establish the LSP. why would we need this command on the Core in the first place?? Regards, Mohamed Hello Mohamed, Well I am not sure why the people you have conferred with are not using it but for me it makes sense to be consistent throughout the core if your intention is for only the PE to PE loopback traffic to be label switched. You have to remember that each device independently allocates and advertises labels for all prefixes learnt via IGP in its RIB. If we do not configure this command on the P routers they will advertise labels for all prefixes in IGP which does not create an issue per say but is not really required either. The reason I responded to your original post was because to me it gave the impression that using this command on the core P routers creates an issue which it does not. Atif ... View more

giuslar wrote: Hello Atif, ten years ago,  C7500 were used as PE nodes and they had a label range 16-50000 thatcould be increased with configuration. for security reasons I would see the use of LDP sessions with MD5 authentication I have seen the introduction of MPLS label filtering for scalability reasons in those times. However, label filtering can be a way to avoid to propagate label binding for unexpected IP prefixes and this can be seen as way to improve safety and protection from human configuration errors. I agree on this Edit: may you provide a link to the feature that you have mentioned Local Label allocation filtering ? thanks a lot I mean mpls ldp avertising-tags for .... not the to option the to option I agree can be a security measure Hope to help Giuseppe Hello Giuseppe, The LLAF feature is documented at: http://www.cisco.com/en/US/docs/ios/mpls/configuration/guide/mp_ldp_alloc_filter.html Atif ... View more

@Guiseppe - This feature is not really used to 'conserve' label space. It is used to control label advertisement be it for security reasons or to ensure only required traffic is label switched. Label conservation is better achieved using the 'Local Label Allocation Filtering' feature (LLAF) using which you can actually do selective label assignments locally. @Mohamed - I do not agree with the statement that the label advertisement control should not be used on the P nodes. It can be used and in fact is used by SPs. As long as you advertise labels from each P node for the PE loopback prefixes you ensure an end-to-end LSP. Atif ... View more

Hello Mahesh, The information is there but in the form of database. I would love to hear from you if you find something closer to what you are looking for. Atif ... View more

mlatosie wrote: Rich, Can you be a bit more specific on what you're trying to achieve? For example if you want to create MPLS VPN yes can still use this topology (the other quiestion is if it makes sense or not, because you will not be able to remonstrate much apart from leaking routes). However to fully simulate MPLS VPN you need: (site1) CE router ----> PE router ----> P router ----> PE router ---> another CE router (site2) Which is a more adequate representation of what would be going on in a real life scenario. You can also built your own MPLS "backbone" by connecting site 1 and site 2  routers with a GRE tunnels (without any interaction from ISP). For practical reasons I have never seen (maybe it does exists) that a ISP would allow you to talk MPLS with their network. Marcin Just to add to what Marcin said if you actually need to run true MPLS between your sites through the SP network you will need to do it over a tunneling mechanism. MPLSoGRE is a possible solution but it's support is platform dependent. There is another implementation called CsC (Carrier Supporting Carrier) which can be used to exchange labels with SP via either LDP or BGP but it is not true MPLS between your sites connected to SP network. If you explain what you are trying to achieve (as Marcin requested) we can probably guide you better. Atif ... View more