Emerging parallel real-time and multimedia applications broaden the range of performance requirements imposed on the interconnection network. This communication typically consists of a mixture of different traffic cla...
详细信息
For performance reasons, client applications often need to influence the implementation strategies of libraries whose services they use. If an object-oriented library contains multiple service classes customized for d...
ISBN:
(纸本)9780897916882
For performance reasons, client applications often need to influence the implementation strategies of libraries whose services they use. If an object-oriented library contains multiple service classes customized for different usage patterns, applications can influence service implementations by instantiating the customized classes that match their needs. However, with many similar service classes, it can be difficult for applications to determine which classes to instantiate. Choosing the wrong class can result in very subtle errors since a customized class might use optimizations that work only over a restricted domain. In this paper, we show how client-side software contracts and exemplar-based class factories can be used to construct customized server objects. By expressing priorities and requirements in contracts, clients can delegate service class selection to the library and thereby avoid implicit dependencies on the library implementation. We have used this approach in the implementation of a real-time database system.
This paper considers the problem of handling schedule disruptions in a distributed manufacturing system. When a cell controller reschedules a manufacturing cell in response to some disruption, it may disrupt the sched...
详细信息
This paper considers the problem of handling schedule disruptions in a distributed manufacturing system. When a cell controller reschedules a manufacturing cell in response to some disruption, it may disrupt the schedule at some other cell, because schedules at different cells may interact. In the approach we propose, a controller at a disrupted cell tries to reschedule in a way which is likely to be least disruptive to other cells' schedules, through negotiation with controllers at other cells. This approach, which we call "polite rescheduling", has the advantage of retaining much of the original schedule, while avoiding wide propagation of the disruption through the rest of the system. Simulation results show that a polite rescheduling algorithm helps isolate disruptions to a small subset of cells.< >
Real-time communication typically consists of guaranteed packets that must satisfy their delivery deadlines and best-effort packets that can tolerate occasional deadline misses for improved average latency. This paper...
详细信息
Real-time communication typically consists of guaranteed packets that must satisfy their delivery deadlines and best-effort packets that can tolerate occasional deadline misses for improved average latency. This paper presents hardware techniques for supporting the coexistence of these two traffic classes in real-time point-to-point networks. A careful selection of routing and switching techniques, coupled with fine-grain arbitration between traffic classes, can allow network adapters to support the diverse performance requirements of best-effort and guaranteed communication. Cycle-level simulations of SPIDER (Scalable Point-to-point Interface DrivER), a network adapter for point-to-point distributed systems, demonstrate the utility of supporting multiple low-level communication policies for different classes of traffic.< >
In this paper, we describe an object-oriented memory-based real-time database system called MDARTS (Multiprocessor Database Architecture for Real-Time Systems). MDARTS is specifically designed to support high-speed re...
详细信息
In this paper, we describe an object-oriented memory-based real-time database system called MDARTS (Multiprocessor Database Architecture for Real-Time Systems). MDARTS is specifically designed to support high-speed real-time applications such as next-generation manufacturing system controllers. MDARTS allows applications to specify their real-time requirements in application code, and during object initialization it attempts to guarantee that these requirements will be met. We have implemented MDARTS on Sun workstations and VME-based multiprocessors and have used our prototype to control an actual manufacturing machine. Our MDARTS prototype can guarantee transaction response times of about 100 microseconds for typical memory-based transactions on VME multiprocessors using 68030 processors.
Modeling fault behaviors such as fault occurrences and active/benign durations is an essential step to the design and evaluation of fault-tolerant controller computers. We use a beta-binomial distribution to model fau...
详细信息
Modeling fault behaviors such as fault occurrences and active/benign durations is an essential step to the design and evaluation of fault-tolerant controller computers. We use a beta-binomial distribution to model fault occurrences both in the presence and in the absence of environmentally-induced (thus common-cause) faults. A multinomial distribution is used to model fault active durations. The proposed model is validated by testing it against the data generated by a simulation program that mimics a common-cause fault environment. The model is then applied to the determination of an optimal time-redundancy recovery method for EMI-induced failures in an N-modular redundant controller computer, demonstrating its utility and power.< >
SPIDER is a network adapter that provides scalable communication support for point-to-point distributed systems. The device exports an efficient interface to the host processor, provides transparent support for depend...
详细信息
SPIDER is a network adapter that provides scalable communication support for point-to-point distributed systems. The device exports an efficient interface to the host processor, provides transparent support for dependable, time-constrained communication, and handles packet routing and switching. The communication support provided by SPIDER exploits concurrency between the independent data channels feeding the point-to-point network, and offers flexible and transparent hardware mechanisms. SPIDER allows the host to exercise fine-grain control over its operation, enabling the latter to monitor and influence data transmission and reception efficiently. In the current implementation, SPIDER interfaces to the Ironics IV-3207, a VMEbus-based 68040 card and will be controlled by x-kernel, a communication executive allowing the flexible composition of communication protocols.< >
Model-based evaluation of reliable distributed and parallel systems is difficult due to the complexity of these systems and the nature of the dependability measures of interest. The complexity creates problems for ana...
详细信息
Model-based evaluation of reliable distributed and parallel systems is difficult due to the complexity of these systems and the nature of the dependability measures of interest. The complexity creates problems for analytical model solution techniques, and the fact that reliability and availability measures are based on rare events makes traditional simulation methods inefficient. Importance sampling is a well-known technique for improving the efficiency of rare event simulations. However, finding an importance sampling strategy that works well in general is a difficult problem. The best strategy for importance sampling depends on the characteristics of the system and the dependability measure of interest. This fact motivated the development of an environment for importance sampling that would support the wide variety of model characteristics and interesting measures. The environment is based on stochastic activity networks, and importance sampling strategies are specified using the new concept of the importance sampling governor. The governor supports dynamic importance sampling strategies by allowing the stochastic elements of the model to be redefined based on the evolution of the simulation. The utility of the new environment is demonstrated by evaluating the unreliability of a highly dependable fault-tolerant unit used in the well-known MARS architecture. The model is non-Markovian, with Weibull distributed failure times and uniformly distributed repair times.< >
暂无评论