一种可分片预留接纳控制算法研究

接纳控制算法的好坏直接影响分布式计算中资源提前预留机制的总体性能.针对现有灵活资源预留接纳控制算法的优缺点，提出了一种可分片预留接纳控制算法.当无法实现固定资源预留时，该算法在保证最大分片间隔的前提下，允许对资源进行分片预留；在各分片中，若存在剩余资源量小于请求预留资源量的时隙，允许用最小资源量进行预留.通过与3种可拓展预留接纳控制算法(缩短持续时间，增大预留带宽(shorten the duration and increase the reserved bandwidth, SDIB)；减小预留带宽，延长持续时间(reduce the reserved bandwidth and extend the duration, RBED)；改变预留的开始时间(change the reserved start time, CST))的对比实验，从接纳率和有效资源利用率方面进行了评估.实验结果表明，可分片预留接纳控制算法能有效减少资源碎片，具有更优的综合性能.

A Fragmentable Admission Control Algorithm for Resource Reservation

Admission control algorithms have direct impact on the overall performance of advance reservation mechanisms in the distributed computing. Based on the advantages and disadvantages of existing flexible admission control algorithms, a fragmentable admission control algorithm for resource reservation is proposed. When the fixed resource reservation cannot be achieved, the algorithm allows reserving resources in fragments with the maximum fragment interval. And when there are slots whose remaining amount of resources is less than the requested amount in a fragment, it allows setting aside the minimum amount of resources. Experiments with other three kinds of extensible admission control algorithms (shorten the duration and increase the reserved bandwidth; reduce the reserved bandwidth and extend the duration; change the start time) show that the fragmentable admission control algorithm for resource reservation can effectively reduce the resource fragmentations, and obtain better performance in acceptance rate and effective resource utilization.