|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ABSTRACT
In modern packet switches, technology limitations may introduce switch configuration delays that are nonnegligible compared with the time required to transmit a single packet. In this paper, we propose a methodology for scheduling of packets, in the context of these technology limitations. If the total tolerable delay through a packet switch is at least on the order of the switch configuration delay, we show that a near 100% utilization of the communication links is possible, while providing strict quality of service guarantees. The main idea is to increase the quantum with which data is scheduled and switched to beyond that of a single packet. This also decreases the rate at which scheduling need to be made, and hence decreases the implementation complexity. The quality of service guarantees we consider are in terms of a service curve. Specifically, we present a framework for the provision of service curves while coping with non-negligible switch configuration delays. REFERENCES
Note: OCR errors may be found in this Reference List extracted from the full text article. ACM has opted to expose the complete List rather than only correct and linked references.
INDEX TERMS
Primary Classification:
Additional Classification:
General Terms:
Keywords:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||