RSDS - Experiences in Building a Rotation-Sensitive Disk Scheduler

Faculty: Tzi-cker Chiueh

Group Members:

• Lan Huang

Project Description:

  The goal of this research is to answer the following question: since most disks does not have complex disk scheduler such as SATF (shortest access time first), can we implement such an intelligent disk scheduler at software level? The challenges for doing this are the following:

1. Can we acquire the accurate disk layout information?
2. Can the disk service time per request be relatively accurate predicted?
3. Is there a simple linear model to model the disk service time per request?

We did extensive micro-benchmarks on various kinds of modern hard drives. The results convinced us that given the complexity of modern hard drive and current limited hard drive interface, it is impossible to have an accurate linear service time model at software level. More details can be found at our technical report.

Technical Report:

• Lan Huang, Tzi-cker Chiueh, "Experiences in Building a Software-Based SATF Disk Scheduler,"   Technical Report ECSL-TR81, revised in July, 2001.

Acknowledgement