This implementation of the firewall enables a high degree of traffic selectability yet avoids the usual performance penalty associated with IP level firewalls. This approach is applicable to high-speed broadband networks, and asynchronous transfer mode (ATM) networks are addressed in particular. Security management is achieved through a new technique of active connection management with authentication. Past approaches to network security involve firewalls providing selection based on packet filtering and application level proxy gateways. IP level firewalling was sufficient for traditional networks but causes a severe performance degradation in high speed broadband environments. The approach described in this paper discusses the use of an FPGA-based front end processor that filters relevant signaling information to the firewall host while at the same time allowing friendly connections to proceed at line speed with no performance degradation.

This paper presents a media access protocol designed for LIGHTING, a WDM testbed currently under construction which has been designed for high-performance supercomputer interconnection. The architecture is based on a dynamically reconfigurable hierarchical WDM network that is being constructed to interconnect a large number of supercomputers and create a distributed shared memory (DSM) environment. This paper describes the network media access protocol based on a single tunable transmitter and single tunable receiver (TT-TR) per node. The protocol exploits the bimodal traffic characteristics of a DSM system. The primary objectives of the protocol design are reduced average latency per packet, support of broadcast/multicast, and support of collisionless communication. The proposed approach is compared to an earlier protocol based on one tunable transmitter and one fixed receiver (TT-FR) per node. The performance of the protocol in terms of average latency and channel utilization is analyzed for varying system characteristics such as number of nodes and channels.

LIGHTNING is a dynamically reconfigurable WDM network testbed project for supercomputer interconnection. This paper describes a hierarchical WDM-based optical network testbed that is being constructed to interconnect a large number of supercomputers and to create a distributed shared memory environment. The objective of the hierarchical architecture is to achieve scalability yet avoiding the requirement of multiple wavelength tunable devices per node. Furthermore, single-hop all-optical communication is achieved: a packet remains in the optical form from source to destination and does not require intermediate routing. The wavelength multiplexed hierarchical structure features wavelength channel re-use at each level, allowing scalability to very large system sizes. It partitions the traffic between different levels of the hierarchy without electronic intervention in a combination of wavelength- and space-division multiplexing. A significant advantage of this approach is its ability to dynamically vary the bandwidth provided to different levels of the hierarchy. Each node in LIGHTNING receives traffic on n channels in an n-level hierarchy, one channel for each level. Each node monitors the traffic intensities on each channel and can detect any temporal or spatial shift in traffic balance. LIGHTNING can dynamically reconfigure to balance the traffic at each level by moving wavelengths associated with each level up or down depending on need. Bandwidth re-allocation is completely decentralized-any node can initiate it, achieving highly fault tolerant system behavior. This paper describes the system architecture, network and memory interface, and the optical devices that have been developed in this project.

Presents an analysis of network performance degradation induced by workload fluctuations. The basic idea of the analysis is based on catastrophe theory. Conventional performance models typically express workload only by using its average properties. However, it may not be possible to predict network performance simply by using average workload properties, since computer networks are nonlinear systems and workload of computer networks usually fluctuates intensely. The interaction between workload fluctuations and the nonlinearity may greatly affect network behavior. The present analysis shows that fluctuations in workload may change the dynamic structure of the system state space and cause the network performance to vary drastically.>

Passive star-coupled optical interconnects with wavelength-division multiplexing hold a strong potential for realizing flexible large-scale multicomputer networks. Virtual point-to-point regular connectivities can be defined and modified in a network with broadcast-select routing via distributed wavelength assignment to the nodes. The system size in this approach is bound by the number of separable wavelength channels. The paper proposes a network class that achieves scalability by grouping the nodes into clusters and employing a separate pair of broadcast and select couplers with each cluster. Interconnecting the clusters according to a regular topology results in a modular architecture with reconfigurable partitions and significantly reduced fiber link density. This approach efficiently combines wavelength-division with direct space interconnection, taking advantage of the properties of each. For the topologies of most significance to parallel computing, the paper studies conflict-free wavelength assignment that maximizes spatial reuse, identifies the valid network partitions, and evaluates the fiber link density and both space and wavelength channel throughput improvement. The results show that cube and shuffle networks take most advantage of the proposed approach in terms of maximizing wavelength reuse.>

Scalable, hierarchical, all-optical, WDM networks for processor interconnection have been previously considered. The paper introduces an access protocol for this type of network which supports a distributed shared memory environment. The objectives of the protocol are reduced average latency per packet, support of broadcast multicast, collisionless communication, and the exploitation of the inherent distributed shared memory traffic characteristics. A hybrid strategy, in terms of channel reservation, is described that trades maximum capacity for reduced communication latency to improve system response. The performance of the protocol is analyzed through analytic and simulation models with varying system parameters such as number of nodes and channels. The performance of the new protocol is compared to a TDM-based protocol and their relative merits and demerits are examined.>

The above paper by L. Bhuyan and D.P. Agrawal (1984) described a hierarchical, all-optical wavelength division multiplexed (WDM) network that is being built to support the communication requirements of a large distributed shared memory system. Dynamically adaptable bandwidth allocation is supported, both within and between levels of the hierarchy, and is highly scalable through wavelength re-use at each hierarchical level. The mixed radix scheme introduced was used for processor numbering, but was not specified. The system was described in term of a time multiplexed access protocol, generalized to the multichannel WDM environment, which has static slot assignment that provides excellent throughput but long latencies due to cycle synchronization. The actual system implementation will use a hybrid WDMA strategy, which is collisionless, provides low latency support, dynamic bandwidth allocation within a hierarchical level, and a fast reliable broadcast.>

This paper introduces a hierarchical optical structure for processor interconnection and evaluates its performance. The architecture is based on wavelength division multiplexing (WDM) which enables multiple multi-access channels to be realized on a single optical fiber. The objective of the hierarchical architecture is to achieve scalability yet avoid the requirement of multiple wavelength tunable devices per node as with the WDM-based hypercube interconnection scheme. Furthermore, single-hop communication is achieved: a packet remains in the optical form from source to destination and does not require cross dimensional intermediate routing. The wavelength multiplexed hierarchical structure features wavelength channel re-use at each level, allowing scalability to very large system sizes. It employs acousto-optic tunable filters in conjunction with passive couplers to partition the traffic between different levels of the hierarchy without electronic intervention. A significant advantage of the proposed structure is its ability to dynamically vary the bandwidth provided to different levels of the hierarchy. The architecture is compared to a wavelength-flat architecture in terms of physical and performance scalability.>

The impact of the wavelength-tunable device switching latency on the performance of media access protocols for star-coupled wavelength-division-multiplexed (WDM) photonic networks is investigated. The effects of a reservation-base protocol (TDMA-C) and two preallocation-based protocols (I-TDMA* and I-SA) are compared. TDMA-C is control-channel based, with one WDM channel allocated to reserve access for data packet transmission on the remaining data channels. I-SA and I-TDMA* are designed for a network where channels are preallocated to the nodes for reception, with each node having a home channel it uses for all data packet receptions. The performance of the protocols is evaluated through discrete-event simulation is terms of network throughput, packet delay, and data channel utilization, with variations in the number of nodes and data channels, packet generation rate, data packet length, optical device switching latencies, and propagation delay.>

Media access control protocols for an optically interconnected star-coupled system with preallocated wavelength-division multiple-access channels are discussed. The photonic network is based on a passive star-coupled configuration in which high topological connectivity is achieved with low complexity and excellent fault tolerance. The channels are preallocated to the nodes with the proposed approach, and each node has a home channel it uses either for data packet transmission or data packet reception. The performance of a generalized random access protocol is compared to an approach based on interleaved time multiplexing. Semi-Markov analytic models are developed to investigate the performance of the two protocols. The analytic models are validated through extensive simulation. The performance is evaluated in terms of network throughput and packet delay with variations in the number of nodes, data channels, and packet generation rate.>

Introduces two hierarchical optical structures for processor interconnection and compares their performance through analytic models and discrete-event simulation. Both architectures are based on wavelength division multiplexing (WDM) which enables multiple multi-access channels to be realized on a single optical fiber. The objective of the hierarchical architectures is to achieve scalability yet avoid the requirement of multiple wavelength tunable devices per node. Furthermore, both hierarchical architectures are single-hop: a packet remains in the optical form from source to destination and does not require cross dimensional intermediate routing.>

This paper compares the performance of two time-division multiplexed media access protocols designed for star-coupled wavelength-division multiplexed (WDM) photonic networks. The TDMA-C protocol is reservation based: one WDM channel is used to reserve access for data packet transmission on the remaining data channels. Control channel access arbitration is achieved through time-division multiplexing, enabling all active nodes the opportunity to transmit once every control cycle. This approach significantly reduces the long synchronization delays typical of time-division multiplexing systems since the control cycle length is proportional to the control packet size rather than the data packet size. Every node in this system has one turnable transmitter and two receivers.>

Media access control protocols for an optically interconnected star-coupled system with pre-allocated WDMA channels are introduced and compared. The photonic network is based on a passive star-coupled WDM–based configuration with high topological connectivity and low complexity. The channels are pre-allocated to the nodes with this approach, where each node has a home channel that it uses either for all data packet transmissions or all data packet receptions. A home channel may be shared if the number of nodes exceeds the number of channels in the system. This approach does not require both tunable transmitters and tunable receivers. The performance of a generalized random access protocol is compared to a protocol based on interleaved time multiplexing. Both protocols are designed to operate in a multiple-channel multiple-access environment and require each node to possess a tunable transmitter and a fixed (or slow tunable) receiver. Semi-markov analytic models are developed to investigate the performance of the two protocols. The analytic models are validated through simulation and performance is evaluated in terms of network throughput and packet delay with variations in system parameters.

A hypercube-based structure in which optical multiple access channels span the dimensional axes is introduced. This severely reduces the required degree, since only one I/O port is required per dimension. However, good performance is maintained through the high-capacity characteristics of optical communication. The reduction in degree is shown to have significant system complexity implications. Four star-coupled configurations are studied as the basis for the optical multiple access channels, three of which exhibit the optical self-routing characteristic. A performance analysis shows that through the integration of agile sources or receivers, and wavelength division multiple access, systems can be developed with significant increases in performance yet at a reduction in communication subsystem complexity.>

A multiprocessor system with a large number of nodes can be built at low cost by combining the recent advances in high capacity channels available through optical fiber communication.A highly fault tolerant system is created with good performance characteristics at a reduction in system complexity.The system capitalizes on the optical selfrouting characteristic of wavelength division multiple access to improve performance and reduce complexity.This paper examines typical optical multiple access channel implementations and shows that the star-coupled approach is superior due to optical power budget considerations.Star-coupled configurations which exhibit the optical self-routing characteristic are then studied.A hypercube based structure is introduced where optical multiple access channels span the dimensional axes.This severely reduces the required degree since only one 1/0 port is required per dimension, and performance is maintained through the high capacity characteristics of optical communication.1

A distributed computer system based on a hypercube topology is proposed, where multiaccess channels spanning all dimensional axes provide processor interconnection. The multiprocessor system can be built at relatively low cost by combining the recent advances in high-capacity channels available through fiber optics, and demand-assignment multiple-access protocols, creating a highly fault-tolerant system with good average packet delay and average distance characteristics, at a significantly reduced cost. A totally distributed system has been achieved in which the packets are self-routing, able to take advantage of the multiple paths of equal length between two modes. The proposed structure gives a highly fault tolerant system, degrading gracefully in the presence of failures.>

An approach to local area network interconnection is presented Which combines the advances in static interconnection topologies, demand assignment multiple access protocols, and the availability of high-bandwidth fiber optic channels to create a cost-effective structure capable of interconnecting a large number of LAN's with heavy traffic. This approach is independent of the protocol implemented at each LAN. The structure is based on a hypercube topology where each vertex of the graph represents a LAN. Multiple access channels spanning all dimensional axes are used in this scheme. This approach is compared to a topology with direct point-to-point connections between all nodes sharing a common axis. Through the development of the degree, diameter, average distance, cost, and average packet delay, we show that using fewer high-capacity channels, a LAN interconnection network with excellent performance characteristics can be constructed, able to support a large number of LAN's with heavy traffic at a significant reduction in cost over the point-to-point case. The resulting structure has many of the desirable characteristics for static interconnection networks such as high fault tolerance, totally distributed packet routing in the interconnection network, low average distance for good performance, and low degree, resulting in low cost. For the total number of required LAN nodes and the expected amount of internode traffic, the structure is optimized for minimum cost.