RP Best practice in Multicast networks

Dieg0MDieg0M Member Posts: 861
Is there a best practice for the placement of the RP in a network topology? I understand the forming of (*,G) generates very little overhead as the RP only starts new sessions but is there a best design practice to optimize performance for other reasons? For example should you put the RP in or out of the multicast flow path and why?

Thank you
Follow my CCDE journey at www.routingnull0.com

Comments

  • networker050184networker050184 Mod Posts: 11,962 Mod
    Depends on the size of the network. An enterprise level network you can probably not worry too much about placement. A core device usually would be in the traffic path and be able to handle the load (traffic and CPU wise) of serving as the RP and makes sense not to have traffic going too out of the way to reach an RP.

    A larger service provider type network I'd go with an anycast RP and put it on the distribution layer to provide some load sharing and redundancy.
    An expert is a man who has made all the mistakes which can be made.
  • Dieg0MDieg0M Member Posts: 861
    And in the case of Bidirectional PIM?
    Follow my CCDE journey at www.routingnull0.com
  • networker050184networker050184 Mod Posts: 11,962 Mod
    I don't have any real world experience with BIDIR-PM other than reading about it so I really can't give you much more than you would find with a google search unfortunately.
    An expert is a man who has made all the mistakes which can be made.
  • srgsrg Member Posts: 140
    An RP will always more or less be in the flow path. More so with BIDIR since you only have (*,G) towards the RP and never join the SPT.
Sign In or Register to comment.