Error detection/correction and fault detection/recovery – Data processing system error or fault handling – Reliability and availability
Reexamination Certificate
1999-05-21
2002-12-17
Baderman, Scott (Department: 2184)
Error detection/correction and fault detection/recovery
Data processing system error or fault handling
Reliability and availability
C709S235000
Reexamination Certificate
active
06496941
ABSTRACT:
TECHNICAL FIELD
This invention relates to communications networks. More particularly, this invention relates to an improved system and method for quickly recovering from failures or error conditions.
BACKGROUND OF THE INVENTION
A telecommunications network transports information from a source to a destination. The source and destination may be in close proximity, such as in an office environment, or thousands of miles apart, such as in a system transmitting credit card transaction data throughout the United States. The information (traffic), which may be, for example, computer data, voice transmissions, or video programming, usually enters and leaves a network at nodes (also termed backbone switches or offices), and is transported through the network via links and nodes. The overall traffic comprises multiple data streams which may be combined in various ways and sent on common links.
Nodes are devices or structures that direct traffic into, out of, and through a network. They can be implemented electronically, mechanically, optically, or in combinations thereof, and are known in the art. Nodes range in complexity from simple switching or relay devices to entire buildings containing thousands of devices and controls. Nodes in a network can be controlled by a central network operations center (“NOC”) and can be programmed with varying degrees of automated traffic-managing capabilities. Links, which may be termed trunks, connect nodes and transmit data between nodes.
A node may become inoperative in a number of ways: for example, power outage, flood or an abnormal amount of messages flooding the network. A link can become inoperative in numerous ways, but most often becomes inoperative as a result of being cut. A network error condition or a network failure is any condition or occurrence that adversely affects the performance of a network or interrupts traffic flow; such a condition may affect only a portion of the network. For example, an error condition may be the failure of a link, a software or control failure, or an overload condition.
Because of the significant volume of traffic typically transported by a network, any disruption in traffic flow can be devastating to large numbers of users transmitting information. The ability to quickly restore network service should a portion of the network become inoperative is of high priority.
A frame relay network is a communications network which transmits data of variable length packets between two points. A frame relay network may accept data in a frame relay format, convert the data to asynchronous transfer mode (“ATM”), transmit the data in ATM form, and convert the data back to a frame relay form when the data leaves the network. ATM uses packets of a fixed length. Thus in such a network a variable length frame entering the network may be broken up into multiple packets of a set length, which are reassembled into the frame when the data leaves the network.
Traffic is routed through a network via a path, a physical or logical route between two points in a network. A path between any two nodes is a route allowing for data transmission between those two nodes; a path may be one link, or may be comprised of multiple links and nodes and other network elements. The length of a path is an indication of the amount of equipment comprising the path; for example, meters of fiber or number of hops (links separated by nodes). A network may transmit data via virtual circuits. A virtual circuit is a path transmitting data between two endpoints in a manner giving the appearance that a dedicated path exists between the two endpoints; in reality any of numerous paths, each path having multiple links and nodes, may be used to connect the two endpoints. For any number of reasons a network may reconfigure a virtual circuit, i.e., change the routing scheme of the virtual circuit.
In one frame relay network, users transmitting data may have a router at a user site for connecting with the frame relay network via an edge vehicle switch (located remotely from the user site) which in turn connects to a node in the network. A user sends data in frame relay form to the network via the router and edge vehicle switch.
The edge vehicle switch converts the data to packets of standard length. The packets are sent through the network via a virtual circuit. An edge vehicle switch connecting to one end of the virtual circuit converts the data to frame relay form and transmits the data to a router located at a user site.
A permanent virtual circuit (“PVC”) is a virtual circuit having a path which is relatively stable over time. In one known network, each PVC is owned by a master node. The master node owning a PVC establishes, monitors, and maintains the PVC and is typically one of the two endpoint nodes for the PVC. Each node is responsible for allocating the capacity of the trunks directly connected to it. Most, if not all, nodes in such a network are both master nodes, owning many PVCs, and via nodes, part of many PVCs owned by other nodes. Establishing a PVC involves finding a path for the PVC. The master node determines a path based on its knowledge of network capacity and transmits requests to numerous potential via nodes in the network. A requested via node responds negatively to a request only if the master node is incorrect as to the trunk capacity allocated by the requested via node, and the trunks for which the requested via node is responsible do not have the capacity to participate in the PVC.
In such a network a failure of a network component, e.g., a node or link, affects multiple PVCs. For example, if one node fails, data cannot flow on the numerous PVCs which use that node as a via node. The affected PVCs must be rerouted: for each PVC the master node owning the PVC must select a set of nodes from the remaining healthy nodes in the network to re-form the PVC. This must be done quickly, and must be done for numerous PVCs, as the failure of even a single node or link may interrupt data transmission for many PVCs.
The reestablishment of a PVC requires the use of network resources such as the processing time of nodes and the communications resources of the network. In certain networks, on the occurrence of relatively small failures, e.g., the failure of two nodes in a 200 node network, master nodes may recover (i.e., reestablish their PVCs and perform other tasks) simultaneously without interfering significantly with each other's recovery. However, on the occurrence of a major disruption, for example, the failure of a majority of the nodes, the load on various network resources from recovering nodes results in interference between nodes trying to reestablish PVCs, which results in inefficiencies delaying overall network recovery.
As part of a node's recovery process, the node queries and receives responses from other nodes to determine whether the other nodes may become via nodes in PVCs owned by the node. Potential via nodes may accept or decline to become part of a PVC based on the capacity of trunks local to the potential via node and on the resource requirements of the PVC. While a potential via node is being queried by one master node, it is unavailable for querying by another master node. Furthermore, when a via node accepts a master node request, it must reconfigure its equipment to become part of that PVC; it is unavailable to respond to other PVC requests during this time. When a node is unable to respond to the PVC request of a second master node because it is responding to the PVC request of a first master node, a collision occurs; the second master node must back-off and attempt the reroute of the entire PVC at a later time. A collision may also occur if a first master node queries a second master node which is busy making a via request of a potential via node. In general, a collision occurs when two objects or devices in a system attempt to access the same resource at the same time, when the resource can service only one object or device.
A collision and the subsequent reroute reattempt waste the resources of both the master node
Eisenberg Martin
Segal Moshe
AT&T Corp.
Baderman Scott
LandOfFree
Network disaster recovery and analysis tool does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Network disaster recovery and analysis tool, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Network disaster recovery and analysis tool will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2980305