Abstract: Managing packets in a shared memory buffer involves linking buffered packets into a linked list in the order that the packets are written into the shared memory buffer, examining the packets in the order of the linked list to determine which packets are intended for available output links, and then dispatching the oldest packet that is intended for an available output link even if the packet is preceded on the linked list by packets that are older but are intended for unavailable output links. Packets are stored with NEXT pointers that establish the linked list and output link identifiers that identify the intended output links of the packets. The head of the linked list is identified by a linked list HEAD pointer. When a packet is dispatched, the packet is removed from the linked list by adjusting the linked list HEAD pointer or the NEXT pointer of the previous packet.

Abstract: Different levels of service are provided to different types of traffic within a single virtual circuit (VC) by converting the traffic from fixed-length cells to variable-length packets, classifying the packets based on information in the packet headers, associating the packets with a VC, and then implementing class-specific enqueuing and dequeuing of the classified packets on a per-VC basis. Classified packets are dequeued from VC-specific and class-specific queues into VC-specific segmentation and re-assembly (SAR) queues according to an algorithm that is a function of traffic class. The dequeuing algorithm determines the level of service that is provided to the different classes of traffic within each VC. Packets are dequeued from the VC-specific SAR queues and converted back to fixed-length cells according to an algorithm that arbitrates among multiple VC-specific SAR queues.

Abstract: A pool of integers is managed by dividing the pool of integers into groups and initializing, in memory, doubly linked lists related to the groups on an as needed basis in response to requests to allocate integers. If an initialized group of integers no longer includes any allocated integers, the doubly linked list related to the group is deleted from memory. Memory locations of the group-specific doubly linked lists are identified by a hash table that includes a unique hash table pointer for each one of the groups. Each element of the doubly linked list includes a next pointer for identifying a next element and a previous pointer for identifying a previous element. A specific free integer can be allocated in an order of one, O(1), operation by indexing into the linked list and utilizing the next and previous pointers of the linked list element to remove the linked list element from the linked list.

Abstract: Managing a position-dependent data set that is stored in a content addressable memory (CAM) array involves identifying an instance of a base position-dependent data set that exists in software, identifying an instance of an edited position-dependent data set that exists in software, determining differences between the instance of the base position-dependent data set that exists in software and the instance of the edited position-dependent data set that exists in software, and incorporating the differences between the instance of the base position-dependent data set that exists in software and the instance of the edited position-dependent data set that exists in software into an instance of said base position-dependent data set that exists in a CAM array.

Abstract: Multiple techniques for providing failure protection in a network node that includes primary and secondary control modules (CMs) are disclosed. One technique involves synchronizing configuration state information between the CMs by executing a configuration command within the primary CM and then executing the same configuration command within the secondary CM only if executing the configuration command within the primary CM does not cause the primary CM to fail. Another technique involves gleaning table entries from hardware-based forwarding tables and using the gleaned table entries to generate a software-based forwarding table in the secondary CM.

Abstract: The flow of packet-based traffic is controlled to meet a desired rate by calculating, as a moving average, a current rate of packet-based traffic on a link, calculating the sum of the error between the calculated current rate and the desired rate, and determining whether or not packets can flow in response to the calculated sum of the error. When the sum of the error between the current rate and the desired rate indicates that the current rate is less than the desired rate, packets are allowed to flow and when the sum of the error indicates that the current rate is greater than the desired rate, packets are not allowed to flow. The magnitude of bursts can also be controlled by artificially controlling the minimum values of the current rate and the sum of the error. The flow control algorithm can be used for rate shaping or rate limiting.

Abstract: Preventing transmission loops in a ring network that utilizes label switching involves generating right side and left side ring packets from an ingress packet and setting right and left TTL values for the packets that prevent the packets from traveling on the same transmission link. Setting the right and left TTL values of the ring packets to values that prevent the packets from traveling on the same link transforms the physical topology of the ring into a logical arc. In an embodiment, the logical arc established by a network node can be dynamically rotated in response to various network conditions to avoid packet transmissions on a particular link of the ring network.

Abstract: A system and a method for managing forwarding table lookups and updates involve maintaining a first forwarding table in a first memory and a second forwarding table in a second memory, and then utilizing the first forwarding table to forward packets while the second forwarding table is updated with current route entries. The second forwarding table is updated in the background and therefore conventional CPU speeds do not cause performance problems. Once the second forwarding table is completely updated, a forwarding table pointer is switched and the second forwarding table is utilized to forward packets while the first forwarding table is updated. Because the second forwarding table has been completely updated in the background, switching the forwarding table pointer causes newly arriving packets to be forwarded according to a forwarding table that is free of inconsistencies.

Abstract: A method and a system for controlling data flow through a network device involve regulating a first flow of network traffic with a first traffic control rule, regulating a second flow of network traffic with a second traffic control rule, and then regulating the passed portion of the two flows with a third traffic control rule that is generic to both the first flow and the second flow. The third traffic control rule regulates the passed portions of the first flow and the second flow on an aggregate basis, such that the total amount of traffic passed by the third traffic control rule does not exceed a predefined bandwidth limit. By regulating the bandwidth consumption of a group of flows with at least two levels of traffic control rules, bandwidth consumption can be simultaneously controlled for narrowly defined flows and more broadly defined flows.

Abstract: Managing a reusable resource, such as a pool of integers, involves a doubly linked list of elements that represents free integers in an integer pool. Each element of the doubly linked list includes a next pointer for identifying a next element and a previous pointer for identifying a previous element. A specific free integer can be allocated from the integer pool in an order of one, O(1), operation by indexing into the linked list and then utilizing the next and previous pointers of the linked list element to remove the linked list element from the linked list. The desired element is removed by adjusting the next pointer of the previous element and the previous pointer of the next linked list element. A non-specific free integer can be allocated from the integer pool in an O(1) operation through a head element that identifies a next free integer in the linked list.

Abstract: Updating a content addressable memory (CAM) involves identifying a new entry that is to be added to the CAM, identifying a free location in the CAM that is the fewest number of prefix levels away from the prefix level of the new entry, moving an existing CAM entry into the free location to create a newly freed location that is a fewer number of prefix levels away from the prefix level of the new entry, repeating the move process until a free location is created at the desired prefix level of the new entry, and then adding the new entry into the newly freed location. The specific algorithm for moving entries to free a location in the desired prefix level is a function of whether the prefix level of the first free location is above or below the prefix level of the new entry.

Abstract: A flexible scheduler in an ATM switch. The scheduler enables each connection to be served fairly according to associated quality of service parameters, while enabling several other features. A connection can be shaped while minimizing additional memory and processing requirements. Specifically, the conformance time of cells of a connections need not be stored when significant backlog exists in the transmission of the cells. The shaping rate can be dynamically varied. Sequence of cells forming a frame are buffered in the ATM switch until the end of frame cell is received. All the cells of a frame are then sent in quick succession.

Abstract: A flexible scheduler in an ATM switch. The scheduler enables each connection to be served fairly according to associated quality of service parameters, while enabling several other features. A connection can be shaped while minimizing additional memory and processing requirements. Specifically, the conformance time of cells of a connections need not be stored when significant backlog exists in the transmission of the cells. The shaping rate can be dynamically varied. Sequence of cells forming a frame are buffered in the ATM switch until the end of frame cell is received. All the cells of a frame are then sent in quick succession.