Abstract
Traditional process schedulers in operating systems control the sharing of the processor resources among processes using a fixed scheduling policy based on the utilization of a computer system such as a real-time or a timesharing system. Since the control over processor allocation is based on a fixed policy, not based on contents or behavior of processes, this can hinder an effective use of a processor or can extend the processing time of a process unnecessarily in some cases. \Ve have already proposed a process scheduling policy, which responds to the behavior of multiple processes of a WWW server, in order to improve the response time of a WWW server. This policy gives any process of a WWW server that is predicted to be a WWW server process handling a text data request from a browser priority over all other processes by moving it to the head of the ready queue where processes waiting for the processor to become available are placed. In this paper, we present the experimental evaluation of our proposed process scheduling policy with regard to the number of simultaneous accesses from browsers and the processor load of the server machine, and explain the results we obtained.
Original language | English |
---|---|
Pages (from-to) | 1752-1761 |
Number of pages | 10 |
Journal | IEICE Transactions on Information and Systems |
Volume | E83-D |
Issue number | 9 |
Publication status | Published - 2000 |
Externally published | Yes |
Keywords
- Execution behavior
- Predict
- Process scheduling
- Response time
- WWW server
ASJC Scopus subject areas
- Software
- Hardware and Architecture
- Computer Vision and Pattern Recognition
- Electrical and Electronic Engineering
- Artificial Intelligence