摘要:
Inter-class schedulers for digital link systems provide high efficiency utilization of limited bandwidth by employing queuing techniques referred to as Statistical Priority Guarantee Queuing (SPGQ) and Generic Cell-Rate Algorithm Priority Guarantee Queuing (GCRA-PGQ). SPGQ “elevates” the priority of otherwise low-priority classes under prescribed circumstances in accordance with a statistical process. The SPGQ scheduler determines whether a number within a range produced by a uniform random number generator lies within a sub-range proportional to the programmed statistical guarantee for a given class. If the number lies within the sub-range associated with a given class, then the priority of that class is elevated to a higher priority when both are eligible to transmit. The GCRA-PGQ scheduler operates as a strict priority mechanism until a class requires bandwidth in excess of a GCRA “window” or threshold for that class. When that occurs, the service priority of that class is temporarily lowered.
摘要:
In Asynchronous Transfer Mode (ATM) communications networks, certain congestion control features, such as Partial Packet Discard (PPD) and Early Packet Discard (EPD), operate on ATM cells that have been segmented from ATM Adaptation Layer #5 (AAL5) frames (packets). AAL5 frames are automatically detected by observing transitions of an indicator in the ATM cell header unique to AAL5 traffic. Automatic detection of AAL5 traffic allows enablement of packet discard techniques such as PPD and EPD.
摘要:
In Asynchronous Transfer Mode (ATM) communications networks, certain congestion control features, such as Partial Packet Discard (PPD) and Early Packet Discard (EPD), operate on ATM cells that have been segmented from ATM Adaptation Layer #5 (AAL5) frames (packets). AAL5 frames are automatically detected by observing transitions of an indicator in the ATM cell header unique to AAL5 traffic. Automatic detection of AAL5 traffic allows enablement of packet-aware cell discarding techniques such as PPD and EPD on a Virtual Channel Connection (VCC) basis, whether in a switch for a VCC or for a Virtual Path Connection (VPC) to which the VCC belongs.
摘要:
In Asynchronous Transfer Mode (ATM) communications networks, certain congestion control features, such as Partial Packet Discard (PPD) and Early Packet Discard (EPD), operate on ATM cells that have been segmented from ATM Adaptation Layer #5 (AAL5) frames (packets). AAL5 frames are automatically detected by observing transitions of an indicator in the ATM cell header unique to AAL5 traffic. Automatic detection of AAL5 traffic allows enablement of packet-aware cell discarding techniques such as PPD and EPD on a Virtual Channel Connection (VCC) basis, whether in a switch for a VCC or for a Virtual Path Connection (VPC) to which the VCC belongs.
摘要:
Methods and systems consistent with the present invention store connection information in a memory of a node in a communications network such that the number of searches for retrieving the connection information is less than a predetermined probe threshold. The node includes a hash table and a connection table in the memory for storing and retrieving information associated with packets, frames, and/or cells in the communications network. Each entry in the hash table includes a connection identifier and, for example, a connection index, and is indexed according to a hash value based on the connection identifier. Each entry in the connection table includes connection state information that is indexed according to a connection index in the hash table. To store connection information in the memory, the node identifies in the hash table a first set of addresses that correspond to a first connection identifier. When the first set of addresses does not correspond to a free entry, the node randomly selects one of the first set of addresses, and identifies a second connection identifier in an entry at the randomly selected address. The node then inserts the second connection identifier at a free one of a second set of addresses corresponding to the second connection identifier and inserts the first connection identifier at the randomly selected address.
摘要:
A data plane architecture enables IP applications, hosted on any processor within a network element, to define steering policies that may be implemented in the data plane. The steering policies enable the applications to define steering policies containing both attributes and actions. The attributes are used to identify the packets and the actions are used to define how the data plane of the network element should handle the packets. The steering policies are implemented in the data plane of the network element by a data path policy steering process. The data path policy steering process operates to implement steering policies defined by applications to cause IP termination to occur at any processor on the network element or optionally at a processor off the network element.
摘要:
A method and system for data packet forwarding is provided. A forwarding information base is stored. The forwarding information base has a first stage with a first lookup key and a second stage with a second lookup key. The first lookup key is a portion of the second lookup key. The first stage is evaluated using the first lookup key forwarding the data packet is forwarded if the first stage evaluation yields a forwarding result. If the first stage evaluation does not yield a forwarding result, the second stage is evaluated using the second lookup key to determine a forwarding result. Such a method and system are suitable for use, among other places, in a network processing unit (“NPU”) in an IPv6 network.
摘要:
Interfaces may be created for applications on a network element without requiring the applications to include code specific for that type of interface. An Interface Management System (IMS) may be included to provide a media abstraction layer between applications and physical media to enable interfaces to be bound to applications without modifying the application to accommodate the interface and without modifying the interface to accommodate the application. The IMS also provides an environment to support distributed interface creation to enable interfaces to be created in a distributed fashion to avoid congestion in the IMS and to facilitate sparing of the IMS and interfaces. The interfaces communicate with applications and with each other through an IFID manager to streamline communications, and to enable the IMS to keep track of the interfaces that have been created on the network element.
摘要:
A system for controlling packet forwarding through a dual-homed connection between first and second network domains, the dual-homed connection including two peer nodes connected to the first and second network domains. A sub-ring network is instantiated in the first network domain, and includes at least two nodes connected in a linear topology between a pair of end-nodes. Each end-node corresponds with a respective one of the peer nodes. A virtual link through the second network domain for conveys traffic of the sub-ring network between the peer nodes, and closes the sub-ring network to define a ring topology. Each of the nodes of the sub-ring network is controlled to forward packets of the sub-ring network in accordance with a ring network routing scheme.
摘要:
A system for controlling packet forwarding through a dual-homed connection between first and second network domains, the dual-homed connection including two peer nodes connected to the first and second network domains. A sub-ring network is instantiated in the first network domain, and includes at least two nodes connected in a linear topology between a pair of end-nodes. Each end-node corresponds with a respective one of the peer nodes. A virtual link through the second network domain for conveys traffic of the sub-ring network between the peer nodes, and closes the sub-ring network to define a ring topology. Each of the nodes of the sub-ring network is controlled to forward packets of the sub-ring network in accordance with a ring network routing scheme.