Shared Trees

• Central point where all receivers attach.

  • Single delivery tree per group. No separate tree per-source.
  • Better use of router resources. No prune-state in off-tree nodes. No unnecessary waste of bandwidth.

• Tree must exist prior to data flow (not data-driven)

• Receivers “explicitly-join” the tree

• Senders need to find paths to reach the shared tree

• Problems:

  • Traffic concentration/bottleneck at center
  • Shared trees not as optimal as source-trees

Previous slide

Next slide

Back to first slide

View graphic version