ATM networks are moving to a state where large production networks are deployed and require a universal, open and efficient ATM network control platform (NCP). The emerging PNNI (Private Network to Network Interface) standard introduces an internetworking architecture which can also be used as an intranetwork interface. However, PNNI fails in the latter due to performance limitations, limited functionality and the lack of open interfaces for functional extensions. OPENET is an open high-performance NCP based on performance and functional enhancements to PNNI. It addresses the issues of scalability, high performance and functionality. OPENET focuses on intranetworking and is fully compatible with PNNI in the internetwork environment. The major novelties of the OPENET architecture compared to PNNI is its focus on network control performance. A particular emphasis is given to the increase of the overall rate of connection handling, to the reduction of the call establishment latency and to the efficient utilization of the network resources. These performance enhancements are achieved by the use of a native ATM distribution tree for utilization updates, lightweight signalling and extensive use of caching and pre-calculation of routes. OPENET also extends PNNI functionality. It utilizes a new signalling paradigm that better supports fast reservation and multicast services, a control communication infrastructure which enables the development of augmented services such as directory, hand-off, billing, security etc. OPENET was implemented by the High-Speed Networking group at Sun Labs and is under operational tests.

Current high-speed packet switching systems, ATM in particular, have large port buffering requirements. The use of highly integrated ASIC technology for implementing high-degree and high-speed switch fabrics is facing a technology mismatch in the sense that today's chip technology does not allow to integrate on-chip the high-speed switching fabric with the large buffering requirements. Consequently, many designs are based on the principles of queueing displacement, i.e., they attempt to move the queueing point off-chip. This is usually done by considerably speeding-up the on-chip switch output ports and placing a second external stage of buffering between the switch fabric and the outgoing link circuitry. Such designs are very popular and are used by many current ATM switch vendors. While such schemes are widely used, no rigourous analysis has so far been offered to evaluate the design trade-offs and to quantify the design points. The model we use to analyze the performance of the above system is a two-node tandem queueing system. The first node in the tandem corresponds to the internal buffer while the second node in the tandem corresponds to the external buffer. It is assumed that the internal buffer is able to transfer c/sub 1/ cells per time unit to the external buffer, while the external buffer is served at a lower rate of c/sub 2/ cells per time unit.

Addresses the problem of designing optimal buffer management policies in shared memory switches when packets already accepted in the switch can be dropped (pushed-out). The goal is to maximize the overall throughput, or equivalently to minimize the overall loss probability in the system. For a system with two output ports, the authors prove that the optimal policy is of pushout with threshold type (POT). The same result holds if the optimality criterion is the weighted sum of the port loss probabilities. For this system, the authors also give an approximate method for the calculation of the optimal threshold, which they conjecture to be asymptotically correct. For the N-ported system, the optimal policy is not known in general, but it is shown that for a symmetric system (equal traffic on all ports) it consists of always accepting arrivals when the buffer is not full, and dropping one from the longest queue to accommodate the new arrival when the buffer is full. Numerical investigations show that under the optimal POT policy the loss probability of a port is insensitive to traffic fluctuations in the other port.

We address the problem of designing optimal buffer management policies in shared memory switches when packets already accepted in the switch can be dropped (pushed-out). Our goal is to maximize the overall throughput, or equivalently to minimize the overall loss probability in the system. For a system with two output ports, we prove that the optimal policy is of push-out with threshold type (POT). The same result holds if the optimality criterion is the weighted sum of the port loss probabilities. For this system, we also give an approximate method for the calculation of the optimal threshold, which we conjecture to be asymptotically correct. For the N-ported system, the optimal policy is not known in general, but we show that for a symmetric system (equal traffic on all ports) it consists of always accepting arrivals when the buffer is not full, and dropping one from the longest queue to accommodate the new arrival when the buffer is full. Numerical results are provided which reveal an interesting and somewhat unexpected phenomenon. While the overall improvement in loss probability of the optimal POT policy over the optimal coordinate-convex policy is not very significant, the loss probability of an individual output port remains approximately constant as the load on the other port varies and the optimal POT policy is applied, a property not shared by the optimal coordinate-convex policy.>

The authors study discrete-time, single server queueing systems, with messages that consist of blocks of consecutive cells. They focus on the model of dispersed cell generation processes which naturally arises in packet switched networks such as ATM. Several important performance measures are considered. These are the message delay process, the maximum delay of a cell in a message and the number of cells in a message whose delays exceed a pre-specified time threshold. The latter two quantities are important for proper design of playback algorithms and time-out mechanisms. They present a new analytical approach that yields efficient recursions for the computation of the probability distribution of each quantity. Numerical examples are provided to compare this distribution with the distribution obtained by using an independence assumption on the cell delays. These examples show that the correlation between cell delays of the same message has a strong effect on each of these quantities.>

Studies buffering policies which provide different loss priorities to packets/cells, while preserving packet ordering (space priority disciplines). These policies are motivated by the possible presence, within the same connection, of packets with different loss probability requirements or guarantees, e.g., voice and video coders or rate control mechanisms. The main contribution of the paper is the identification and evaluation of buffering policies which preserve packet ordering and guarantee high priority packets performance (loss probability), irrespective of the traffic intensity and arrival patterns of low priority packets. Such policies are termed protective policies. The need for such policies arises from the difficulty to accurately characterize and size low priority traffic, which can generate large and unpredictable traffic variations over short periods of time. The authors review previously proposed buffer admission policies and determine if they satisfy such "protection" requirements. Furthermore, they also identify and design new policies, which for a given level of protection maximize low priority throughput.>

Buffering policies that provide different loss priorities to packets/cells with no change in packet ordering (space priority disciplines) are studied. These policies are motivated by the possible presence, within the same connection, of packets with different loss probability requirements or guarantees. Examples of such applications are voice and video coders that generate information of unequal importance, and rate control mechanisms that mark excess traffic with a low priority rate violation tag. The focus is on the identification and evaluation of buffering policies that can guarantee performance, i.e. loss probability, to high priority packets irrespective of the traffic intensity and arrival patterns of low priority packets, while preserving the original ordering among packets. Such policies are termed protective policies.>

A family of queues for which the service time B/sub n/ of customer n depends on the interarrival time I/sub n/ between customers n-1 and n and the random variables I/sub n/ and B/sub n/ exhibit a proportionality relation is studied. In particular, the focus is on dependencies that arise naturally in communication systems, where the finite speed of the communication links constrains the amount of data that can be received in a given time interval. The simple case of a deterministic proportionality relation between the service time of a customer and its preceding interarrival time is considered and extended to allow the addition of an independent, generally distributed overhead to the service time. Several models that capture the on-off behavior of communication links in packet networks are then addressed. In all cases, expressions for the delay experienced by a packet in the system and illustrative numerical examples are provided.>

A family of queuing systems in which the interarrival time I/sub n+1/ between customers n and n+1 depends on the service time B/sub n/ of customer n is considered. Specifically, cases where the dependency between I/sub n+1/ and B/sub n/ is a proportionally relation and B/sub n/ is an exponentially distributed random variable is considered. Such dependencies arise in the context of packet-switched networks from employing rate policing functions which regulate the amount of data that can arrive at a link within any given time interval. The models developed and the associated solutions are, however, of independent interest and potentially applicable to other environments. Several scenarios that consist of adding an independent random variable to the interarrival time, allowing the proportionality to be random, and the combination of the two are considered. Numerical results are compared to those for an equivalent system without dependencies.>

The authors study the message queuing delays in a node of a communication system, where a message consists of a block of consecutive packets. Two types of message generation process are distinguished. The message can be generated as a batch or it can be dispersed over time. The authors focus on the dispersed generation model. The main difficulty in the analysis is due to the correlation between the system states observed by different packets of the same message. A technique for analyzing the message delay in such systems for different arrival models is introduced, and it is shown that the correlation has a strong effect on the performance of the system. For an M/M/1 system with variable size messages, an explicit expression for the Laplace-Stieltjes transform (LST) of the message delay is obtained. It is shown that the commonly used independence assumption can lead to wrong conclusions.>

A family of queues is studied for which the service time B/sub n/ of customer n and the interarrival time I/sub n/ between customers n-1 and n exhibit some sort of proportionality. The focus is on dependencies that arise naturally in the context of communication systems, where the finite speed of the communication links constrains the amount of data that can be received in a given time interval. The simple case of a deterministic proportionality relation between the service time of a customer and its preceding interarrival time is considered. This is extended to allow the addition of an independent, generally distributed overhead to the service time of each customer. Several models that capture the ON-OFF behavior of communication links in packet networks are considered. In all cases, expressions for the delay experienced by a packet in the system are provided. Numerical examples illustrate the impact of dependencies through comparison with less accurate models. The results should be of relevance to environments other than communication as well.>

The packet loss process in a single-server queueing system with a finite buffer capacity is analyzed. The model used addresses the packet loss probabilities for packets within a block of a consecutive sequence of packets. An analytical approach is presented that yields efficient recursions for the computation of the distribution of the number of lost packets within a block of packets of fixed or variable size for several arrival models and several numbers of sessions. Numerical examples are provided to compare the distribution obtained with that obtained using the independence assumption to compute the loss probabilities of packets within a block. The results show that forward error correction schemes become less efficient due to the bursty nature of the packet loss processes; real-time traffic might be more sensitive to network congestion than was previously assumed; and the retransmission probability of ATM messages has been overestimated by the use of the independence assumption.>

An efficient recursive computation methodology is introduced to obtain the exact distribution of the number of lost packets in a block of packet arrivals of a given size for different arrival models and a different number of sessions. The exact distribution is compared with the distribution obtained from an independence assumption on the loss probability of packets. Numerical examples are provided to show that the exact distribution may be worse than the distribution obtained under the independence assumption for applications such as forward error correction or better for applications such as straight message retransmission.>

A class of discrete-time priority queueing systems with Markov modulated arrivals is considered. In these systems, N queues are served by a single server according to priorities that are preassigned to the queues. Packet arrivals are modeled as discrete-time batch processes with a distribution that depends on the state of an independent common two-state Markov chain. This allows coverage of a wide range of applications in computer and communication systems when the parameters of the arrival processes are not fixed in time but vary according to the state of the underlying Markov chain. The steady-state joint generating functions of the queue length distributions of this class of systems are derived. From these, moments of the queue lengths as well as average time delays can be obtained. A numerical example provides some insight into the behavior of such systems. Also, the effect of the transition rate between the states of the modulating Markov chain on the average time delay in the system is investigated for different patterns of loads on the queues of the system.>