Vol II lab 1 Task 7.2

Hi all,

   For the above task in configuring R2 in the policy-map applied in the output direction towards the PE, do we need to use the QoS_group?? can't we match on the experimental bits in the bottom label of the packet??

thanks alot



  • Hi,

    I have had the same questions as to why the set QOS group and match QOS group are part of the solution.  When you do your initial set at the ingress of the provider cloud you use the set mpls experimental imposition X which, as per cisco's documentation "Specifies the experimental-bit value on IP to Multiprotocol Label Switching (MPLS) or MPLS input in all newly imposed labels."  Which I interpret as setting the MPLS exp value on both labels imposed at this point.  So when traveling thru the P network the top-most label is inspected and acted upon until it gets to the next to last hop in which the top label is popped, thus leaving a single VPN label which still has the experimental value of X in which can be matched.  This works just fine without using the QOS group and just keeping the same policy on the next to last hop router as all P others. 

    You can see this at work with the debug mpls packet output.  The output below is a single icmp packet at the penultimate host router in the provider network.  You can see that mpls packet is recieved with the topmost label is 17 and has an exp value of 5 and the VPN label is 19 and has an exp value of 5.  When the topmost label is popped and the packet is sent out, you will be left with the VPN label exp value 5.

    *Jan  9 15:04:33: MPLS turbo: Fa1/1: rx: Len 122 Stack {17 5 251} {19 5 251} - ipv4 data
    *Jan  9 15:04:33: MPLS turbo: Fa1/0: tx: Len 118 Stack {19 5 250} - ipv4 data  

    Possibly the solution is trying to point out the fact penultimate hop popping is taking place. 

    That is where I have questions as well.  Does every label in the stack always have the same exp bit?  I tried to manipulate the top most label in the P network in my scenario and while it modified the top label it also modified the VPN label's exp bit.  I havent found any documentation that specifically states either way, but all my testing shows that the entire label stack has the same exp value throughout and changing one changes all.  Anyone know of any clear documentation?  Maybe different IOS version behave differently?

    Best Regards,



  • Thanks man to your reply, I also think that the Cisco IOS applies the same exp bit value to all the labels in the stack....so for this task we can macth simply on the exp bits value..

    waiting the reply about whether it is IOS dependent or not??



Sign In or Register to comment.