The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2001-Sep> msg00329



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Questions / Comments on MPLS-TE MIB

  • From: Thomas Nadeau <tnadeau@cisco.com>
  • Date: Mon, 24 Sep 2001 13:21:46 -0400
  • Cc: mpls@UU.NET


>         Hi! Here are few questions/comments related to MPLS-TE MIB.
>         Your help will be appreciated.
>
>         Thanks,
>         sanjay
>
>         1. mplsTunnelInstancePriority, is significant only in the context
>         of a Tunnel (group of tunnel instances).
>
>         Therefore, according to the MPLS-TE model,load sharing and
>         failover(primay-backup) mechanisms are valid ONLY within a
>         Tunnel and not across Tunnels.
>
>         Is this the correct interpretation of the MIB ?

         Yes, the tunnelInstancePriority is significant only
within the tunnel instances using the same first index of
the tunnelEntry.

>         2. Currently, the MPLS-TE MIB does not allow the user to
>         specify the traffic class (ex. DiffServ class) associated with
>         the Tunnel Instances.
>
>         Although, one can define new MIBs to add the above mentioned
>         feature, in my opinion, extending the existing (pre-RFC) MPLS-TE
>         BIB, will significantly enhance the usability of the MIB.
>
>         Can/Should this be done?

         This has already been done. The DiffServ Packet
Classifier MIB can be used to handle the DiffServ
marking/matching operation. Once this operation is
done, a packet is ultimately dropped or forwarded. If
it is forwarded, it should arrive at the MPLS FTN MIB.
The MPLS-FTN-MIB allows one to associate incoming traffic
with a tunnel interface.

>         3. If you add the traffic class support to the mplsTunnelTable,
>         should this trafficClass be an (additional) index to the tunnel
>table ?
>
>         i.e. Should all the instances of a traffic tunnel have the same
>         traffic class ?

         I think that is implementation dependent. However, I would
think that which ever LSP (instance) of that tunnel is used to carry
that traffic should probably be specified in such a way as to satisfy
the needs of the traffic, but I am not sure that is a hard
requirement.

         --Tom