Multiplex communications – Data flow congestion prevention or control – Control of data admission to the network
Reexamination Certificate
1999-02-04
2004-04-13
Pham, Chi (Department: 2667)
Multiplex communications
Data flow congestion prevention or control
Control of data admission to the network
C370S358000
Reexamination Certificate
active
06721271
ABSTRACT:
TECHNICAL FIELD
This invention relates generally to the field of data packet switching and, in particular, to a rate-controlled very high-capacity switching node adapted to transfer data packets of variable size.
BACKGROUND OF THE INVENTION
There is a need for data networks adapted to transfer data in packets of varying and arbitrary size. A principal design objective for switching nodes in such networks is to realize a switch of scalable capacity and controlled service quality. A popular approach to achieving this objective is to build networks based on the Asynchronous Transfer Mode (ATM) protocol. ATM protocol has, in fact, succeeded in facilitating the construction of very-high capacity switches and in providing effective means of service-quality control by enabling the enforcement of data transfer rate control. ATM, however, is only adapted to switch packets, referred to as cells, of a 53-byte fixed cell length. Switching cells of fixed size rather than packets of variable size simplifies the design of switches.
ATM switches are cell-synchronous switches and are somewhat simpler to build and grow than a switch adapted to switch variable-sized packets. However, there is a disadvantage in using a network that operates under a protocol which accommodates only fixed-size cells. When variable sized packets are transferred through such a network, the packets must be deconstructed at a network edge device and packed into an appropriate number of cells. In the process of packing the contents of the packets into cells, a proportion of the cells is underutilized and must be padded with null data. Consequently, a proportion of the transport capacity of the network is wasted because of the partially filled cells. Another disadvantage of transferring variable-sized packets in cells is that when a single cell is lost from a packet, the entire packet must be discarded but this is undetectable until all the remaining cells reach a destination edge of the network. When a cell belonging to a given packet is lost, the remaining cells are unknowingly transferred on to the destination edge of the ATM network, only to be discarded there because the packet is incomplete. If a packet is transferred as a single entity the entire packet may be lost at a point of congestion, but further downstream consumption of network resources is avoided.
A node adapted to switch variable-sized packets may be more complex than a node adapted to switch fixed-sized packets (cells). However, the cost resulting from the extra complexity is more than offset by the increased efficiency gained in the network. Connections can be established in a network using variable-sized packet switches, and the connections can be rate-controlled on either a hop-by-hop or end-to-end basis, so that sufficient transfer capacity can be reserved to satisfy service quality requirements. One network architecture for achieving this is a network that employs the Universal Transfer Mode (UTM) protocol, described in detail in U.S. patent application Ser. No. 09/132,465 to Beshai, filed Aug. 11, 1998. UTM retains many of the desirable features of ATM but adds more flexibility, such as the option to mix connection-based and connectionless traffic, yet significantly simplifies connection-setup, routing, and flow-control functions. The simplicity of connection-setup facilitates the use of adaptive means of admission control. Admission control is often based on traffic descriptors which are difficult, if not impossible, to determine with a reasonable degree of accuracy. An alternative is to use adaptive admission control that is based on monitoring the network traffic and requesting a change in transfer allocations on the basis of both the traffic load and class of service distinctions. The traffic at each egress module is sorted according to destination and class of service. The packets thus sorted are stored in separate logical buffers (usually sharing the same physical buffer) and the occupancies of the buffers are used to determine whether the capacity of a connection should be modified.
In a network shared by a variety of users, class of service distinctions are required to regulate traffic flow across the network. Several traffic types may share the network, each traffic type specifying its own performance objectives. Enforcing class of service distinctions within a switching node adds another dimension that can potentially complicate the design of the switch.
A high capacity switch is commonly constructed as a multi-stage, usually three-stage, architecture in which ingress modules communicate with egress modules through a switch core stage. The transfer of data from the ingress modules to the egress modules must be carefully coordinated to prevent contention and maximize the throughput of the switch. Within the switch, the control may be distributed or centralized. A centralized controller must receive traffic state information from each of the N ingress modules. Each ingress module reports the volume of waiting traffic destined to each of N egress modules. The centralized controller therefore receives N
2
elements of traffic information with each update. If, in addition, the controller is made aware of the class of service distinctions among the waiting traffic, the number of elements of traffic information increases accordingly. Increasing the number of elements of traffic information increases the number of control variables and results in increasing the computational effort required to allocate the ingress/egress capacity and to schedule its usage. It is therefore desirable to keep the centralized controller unaware of class of service distinctions while providing a means of taking the class of service distinctions into account during the ingress/egress transfer control process.
A high capacity ATM switch which uses a space switch core to interconnect ingress modules to egress modules is described in U.S. Pat. No. 5,475,679 which issued to Münter on Dec. 12, 1995. The controller of the switch coordinates the transfer of bursts of ATM cells between the ingress modules and the egress modules. One of the limitations of the space switch architecture, whether applied to TDM or ATM, is the necessity to arbitrate among a multiplicity of ingress/egress connection attempts.
This limitation can be removed by spatial disengagement using a rotator-based switch architecture. In the rotator-based switch architecture, the space switch core is replaced by a bank of independent memories that connect to the ingress modules of the switch through an ingress rotator. Traffic is transferred to the egress modules of the switch through an egress rotator. The two rotators are synchronized. A detailed description of the rotator switch architecture is provided in U.S. Pat. No. 5,745,486 that issued to Beshai et al. on Apr. 28, 1998.
The rotator switch architecture described by Beshai et al. works well for fixed length packet protocols such as asynchronous transfer mode (ATM). It is not adapted for use with variable sized packets, however. Consequently, there is a need for a switch that can efficiently transfer variable sized packets. To be commercially viable, the switch must also be adapted to operate in an environment that supports multiple classes of service and is rate-regulated to ensure a committed quality of service.
OBJECTS OF THE INVENTION
It is therefore an object of the invention to provide a method and an apparatus for switching variable sized packets at a controlled rate determined by traffic class of service and destination.
It is another object of the invention to provide a rate-controlled, variable-sized packet switch having a very high capacity.
It is a further object of the invention to provide a rate-controlled, variable-sized packet switch in which a core controller for the switch need not be aware of class-of-service distinctions.
It is yet another object of the invention to provide a variable-sized packet switch in which the variable-sized packets are segmented on ingress into fixed size segments, data in a last segment in each packet being
Beshai Maged E.
Munter Ernst A.
(Ogilvy Renault)
Boakye Alexander O.
Nortel Networks Limited
Pham Chi
Wood Max R.
LandOfFree
Rate-controlled multi-class high-capacity packet switch does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Rate-controlled multi-class high-capacity packet switch, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Rate-controlled multi-class high-capacity packet switch will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3207281