Event Sequence

Until this point, the reservation event sequence described in this document has been from the application and network perspective. This section deals with the reservation process in terms of software events and processes. We will look at two scenarios. For the first scenario, network resources are sufficient and the reservation is successful. In the second scenario, the reservation is not granted due to insufficient resources along the data path.

Scenario A: Resource reservation sequence for a scenario with sufficient network resources

  1. When a new TCP/UDP session is established, the Application-RSVP Interface process sends a request to the local RSVP process to begin RSVP processing for the session.
  2. Upon receiving the request, the local RSVP process sends RSVP signaling Path messages to the receiver.
  3. Path messages are forwarded on a hop-by-hop basis. The Path message creates a Path state on each RSVP-aware intermediate node. The Path state contains source and destination information as well as the IP address of the previous node in the direction of data traffic.
  4. When the Path message arrives at the receiver’s RSVP process, the information contained in the Path message is forwarded to the Application-RSVP interface process.
  5. The Application-RSVP interface process decides if the receiver will request a reservation for the traffic described in the received Path message. If the reservation is to be made, it contacts the local RSVP process.
  6. The receiver RSVP process creates a reservation state and builds the reservation message. This reservation message is then sent to the sender with the destination address of the previous hop in the data path (as learned from a corresponding Path state).
  7. The reservation message is forwarded to the RSVP process at each intermediate router. The RSVP process then makes a reservation request to the local traffic control process and if there are sufficient resources available, the traffic control process grants the reservation and makes appropriate changes to the packet classifying and packet scheduling processes.
  8. If the reservation is granted, the reservation message is forwarded to the sender. Otherwise, a reservation error message is sent to the receiver (see Scenario B).
  9. Steps 6 to 8 are repeated at each intermediate RSVP-aware node.
  10. Once the sender receives the reservation message, it sends an Resv Confirmation message informing the receiver that the reservation has been successful.
  11. The reservation is now established and data traffic will use the reserved resources. Note that in the Riverbed Modeler implementation, the application may start sending data before the reservation is fully established.

Scenario B: Resource reservation sequence for a scenario with insufficient network resources

  1. Steps 1-7 as in Scenario A.

  2. If some intermediate node has insufficient resources, a reservation error message is created and sent to the receiver.

  3. When an intermediate node receives a reservation error message, a Blockade state is created. Establishing this state modifies the merging process in a node so that smaller reservation requests are not blocked by the unsuccessful request.

  4. When the reservation error message reaches the receiver, the local RSVP process forwards the information from the message to the RSVP-Application interface process.

  5. The application interface process decides if another reservation attempt should be made.

    If another reservation attempt is to be made, a request is sent to the local RSVP process to re-try the reservation. This may be done either immediately after receiving Resv Error message, or after some time. The part of already established reservation will be left in place. Otherwise, the reservation is torn down, and data will use the best-effort service.