Multiplex communications – Fault recovery
Reexamination Certificate
1999-03-25
2003-03-25
Chin, Wellington (Department: 2664)
Multiplex communications
Fault recovery
C370S222000, C370S223000, C370S241100
Reexamination Certificate
active
06538987
ABSTRACT:
BACKGROUND OF THE INVENTION
One of the advantages of a ring-based network is that the traffic between two nodes on the ring can be re-routed over a predetermined secondary route, if a failure should occur in a primary route. An example of such a network is a SONET ring, with predefined primary and secondary, or working and protection, routes between the nodes on the ring. The routes may be over redundant rings, which pass traffic simultaneously in opposite directions. Such a system is commonly referred to as a “unidirectional ring.”
When a failure or a significant degradation in, for example, the primary route, is detected on a SONET ring, the system must automatically re-route, or switch, affected traffic from the primary route to the secondary route. The re-routing, which is commonly referred to as “protection switching,” is performed in unidirectional systems by the destination nodes, that is, by the nodes that terminate the traffic or route the traffic off of the ring to a user or another network. In the example, the destination nodes switch from receiving the affected traffic over the primary route to receiving the traffic over the secondary route. The specifications for many SONET ring based networks require that protection switching be accomplished in 50 msec or less.
Recently, SONET rings have been incorporated into ATM systems. In these systems, ATM cells and frames are routed over the ring in virtual circuits. The virtual circuits that span the same nodes are bundled into virtual paths, and the rings are thus referred to as “virtual path rings.” The virtual paths in these virtual path rings are pre-established, in the sense that each virtual path is assigned a fixed amount of the ring bandwidth based, for example, on the service contracts of the associated users. A virtual circuit is included in a virtual path by allocating to it a fixed portion of the path bandwidth. If there is not enough available bandwidth within the virtual path for the virtual circuit, the system refrains from setting up the circuit, even if there is bandwidth otherwise available on the ring. Accordingly, the virtual path rings may not be able to accommodate bursty traffic, such as the traffic from a local area network, or “LAN.” For a more detailed explanation, refer to Bellcore Standard GR-2837.
In known prior systems, the virtual circuits are set up over the selected ring, which in the example is the primary ring. Each destination node includes a primary ring interface that is configured to forward the cells received over the virtual circuits to destination ports in accordance with stored routing information. The destination node discards the corresponding traffic received over the non-selected ring, since the virtual circuits are not terminated on the interface associated with that ring.
Intermediate nodes manage traffic on a virtual path basis, without reference to virtual circuit routing information. The intermediate nodes thus pass the virtual path traffic received on the primary ring to successive nodes on the primary ring and the virtual path traffic received on the secondary ring to successive nodes on the secondary ring, regardless of which ring is ultimately chosen as the selected route by the destination node.
A decision to perform protection switching on the virtual path ring is made on the basis of the virtual paths, and necessarily affects all of the virtual circuits included in the switched virtual paths. The destination nodes implement the switch by individually tearing down the hundreds or thousands of affected virtual circuits set-up on the interface to the previously selected route, and setting up new virtual circuits on the interface to the previously non-selected route. If the SONET ring is not unidirectional, the source nodes must also switch from sending the affected virtual circuit traffic over the previously selected ring to sending the traffic over the previously non-selected ring. The source nodes must thus similarly tear down the hundreds or thousands of virtual circuits from one ring interface and set up new ones on the interface to the other ring.
If the number of affected virtual circuits is relatively large, a given destination node may not satisfy the 50 msec time limit for protection switching. The protection switching time limit may thus limit the number of virtual circuits that can be included in a given virtual path. What is needed is a mechanism that rapidly performs protection switching regardless of the number of affected virtual circuits.
There are essentially two events that trigger protection switching, namely, the failure of a selected path or the degradation of the path. When the path fails, traffic is no longer provided to the destination node over the path. When the path is degraded, the traffic over the path becomes corrupted.
The SONET ring specifications require that intermediate nodes detect a path failure by determining when a node interface on the path is no longer operational. See, Bellcore Standard GR-2980. The intermediate nodes then send appropriate OAM cells to the affected destination nodes, to notify them of the path failure. If the ring is not unidirectional, the intermediate nodes also send OAM cells to the affected source nodes. In response to the OAM cells, the affected destination nodes and, as appropriate, the source nodes, perform protection switching.
The failure detection mechanism works well when the path failure is caused by a failure in the transmission medium, such as a broken cable. It does not, however, work well when the path failure is the result of a node no longer forwarding certain cells over the selected ring because of, for example, an erroneous routing table configuration or a hardware problem. The failure to forward the cells must instead be detected by the intended destination node, when it no longer receives cells. Such failures are not readily detected in traffic that is bursty, such as LAN traffic, because the time intervals between the cells vary in length. Accordingly, there may be a relatively long delay, and thus, a loss of cells, before the destination node detects the path failure. What is needed is a mechanism for the nodes to detect path failure without significant delay, even in bursty traffic.
The SONET ring specifications also require that nodes notify “upstream” nodes of degradation in the path, that is, degradation in the transmission facilities on the path. See, Bellcore Standard GR-2980. The specifications do not, however, state how the nodes determine that a path has become sufficiently degraded, that is, how the nodes determine an accumulated error rate for the path.
Errors in the cell body are detected when the cell is processed at the destination node in an ATM segmentation and reassembly layer, which is removed from the SONET transport layer at which the protection switching decisions are made. There is thus a delay in providing error information from the ATM segmentation layer to the SONET transport layer. Also, cells with damaged headers are discarded along the route and the destination nodes are not typically notified of this type of error.
A system may use OAM cells for monitoring virtual path performance, as set forth in ITUI.610(3/93)B-ISDN Operations and Principles. The performance monitoring OAM cells are injected at periodic intervals into the cell streams on the virtual paths. Each performance monitoring OAM cell includes a count of the cells transmitted after the transmission of the previous associated performance monitoring OAM cell, and summary cell parity information for the counted cells. The count and parity information in the performance monitoring OAM cells are used to determine the number of cells lost over the primary and secondary paths and also associated error bit rates. If the cell losses over the primary path far exceed the cell losses over the secondary path, or there are substantially higher error bit rates associated with primary path then with the secondary path, a protection switch may be initiated.
In order to use the performance monitoring OAM cells, the
Cedrone Dan
Dray Walter
Klein Ed
Ma Ming-Teh
Regan James
Chin Wellington
Ciena Corporation
Sheehan Patricia A.
Soltz David L.
Tran Maikhanh
LandOfFree
Rapid ring protection switching system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Rapid ring protection switching system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Rapid ring protection switching system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3080494