End-to-End Connections

This section analyzes the steps involved in the creation of an end-to-end connection. As you know, the multiservice switching architecture is based on the separation of the switching plane and the control plane. A general three-node network might look like Figure 2-11.

Figure 2-11. Multiservice Switching Network: End-to-End Connections

Based on the networking protocol running in the controllers (at the networking protocol layer), the VSI master instructs the controlled switch to set up the cross-connects that are the segments of the end-to-end connections. The controlled switch does not know the end-to-end connection information or topology, only the local cross-connect.

If the controllers are MPLS controllers, the networking protocol is LDP and an IGP, such as OSPF or IS-IS. The label distribution mechanism is Downstream on Demand, and will be discussed in detail in chapters 4 "Introduction to Multiprotocol Label Switching" and 6 "MPLS Implementation and Configuration." Figure 2-12 shows a multiservice switching MPLS network.

Figure 2-12. Multiservice Switching Network: MPLS Example

NOTE

The label distribution technique used in ATM LSR has some variations that are detailed in Chapters 4 and 6. For ATM-LSRs that are not VC-merge-capable, the distribution mechanism must be Downstream on Demand, and optionally is the same for ATM LSRs that support VC-merge. Downstream on Demand has two control modes: independent (optimistic label allocation) and ordered (conservative label allocation). These are defined in RFC 3031 ("Multiprotocol Label Switching Architecture") and RFC 3036 ("LDP Specification"). These variations determine how end-to-end connections are signaled and set up, and thus at which point the cross-connects are installed in the controlled switches.

NOTE

Here is the definition of VC-merge in RFC 3035 ("MPLS Using LDP and ATM VC Switching"), Section 3: "VC-merge is the process by which a switch receives cells on several incoming VCIs and transmits them on a single outgoing VCI without causing the cells of different AAL5 PDUs to become interleaved."

NOTE

A VPN MPLS network also has edge-to-edge communication. The PE (Provider Edge) devices have an iBGP neighboring relationship.

Let's explore a complete example of how a controller can set up and tear down a connection segment of an end-to-end connection in a multiservice switching PNNI network. See Figure 2-13.

Figure 2-13. Multiservice Switching Network: PNNI Example

The first part of the example details the steps that occur in a Connection SETUP:

Step 1.	Router A sends a SETUP message to router B to establish an SVC.
Step 2.	The controller in the first node receives the SETUP message and sends a CALL PROCEEDING back to the router.
Step 3.	The controller requests a connection segment using VSI to be set up (reserved) at the local controlled switch.
Step 4.	The controller sends the SETUP signaling message to the next node.
Step 5.	The VSI slave in the local controlled switch parses the VSI message from Step 3, checks the CAC, allocates resources (message execution), and sends the response back to the controller (message responding). The response includes updated loading information for the interface.
Step 6.	Router B receives the SETUP message and replies with a CALL PROCEEDING.
Step 7.	Router B sends a CONNECT, and the CONNECT message reaches the controller in the first node.
Step 8.	The controller sends a CONNECT message to router A.
Step 9.	The controller requests the connection segment to be activated (committed) using VSI at the local switch.
Step 10.	The local switch configures the hardware to enable data flow and sends an acknowledgment back to the controller using VSI.

The preceding is a simplified example from a Q.2931 point of view, trying to emphasize VSI messaging in every step. The details of PNNI signaling will be covered in Chapters 8 "PNNI Explained" and 10 "PNNI Implementation and Provision."

The second part of the example describes the stages in a Connection RELEASE process:

Step 1.	The controlled switch detects a failure from the hardware on a specific interface, such as a loss of signal (LOS).
Step 2.	The controlled switch notifies the controller of the logical interface failure using VSI.
Step 3.	The controller sends CALL RELEASE messages for every call on that interface to other controllers and routers.
Step 4.	The controller requests that the cross-connects be removed from the local switch.
Step 5.	The controlled switch removes connection legs and releases resources. The VSI slave sends a connection request response VSI message back to the controller with updated loading information for the corresponding interfaces.