Performance modeling of magnetic disks allows the prediction of the disk service time which is useful for on-line decision support for soft real-time applications. In this paper, we propose a new performance model of disk access time to estimate the bounded disk service time. Our proposed model focuses on modeling the head positioning time to fully utilize the disk I/O bandwidth by exploiting the geometric layout of the disk. The experimental results show that our proposed model can estimate the disk service time with less than 10% error on average.
|Title of host publication||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|Editors||Jing Chen, Seongsoo Hong|
|Number of pages||13|
|State||Published - 2004|
|Name||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|