Abstract

Applications with real-time traffic, such as video and voice, require quality-of-service (QoS) guarantees, such as bounded end-to-end delays and bounded cell loss probabilities. In order to provide the QoS guarantees for each connection in ATM networks, link scheduling to prioritize the transmission of queued cells is desirable. A hardware priority queue is then necessary since fast cell switching is hard to be realized in software. In addition, the queue has to be scalable with respect to the number of cells and the number of priority levels. A failure in the priority queue, however, will jeopardize the QoS guarantees for time-critical cells. Moreover, the errors occurred in the queue are unlikely to be detected at the destination. In this paper, we present a fast and scalable priority queue architecture for ATM switches. The queue can reconfigure itself once an error is detected, and thus it will continue normal operation even in the event of a failure.

Author Keywords: Priority queues; Scalable; Fault; Error detection; Reconfiguration

Article Outline

1. Introduction

2. A scalable priority queue architecture

3. A reconfigurable priority queue

4. Recoverability of comparison errors

5. Conclusions

References