Disk Logging Algorithms and Performance Analysis Based on Petri Net Models

ZHOU Feng1,  LIN Chuang1,  ZHENG Wei-min1

1(Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China);

E-mail: zhoufeng00@mails.tsinghua.edu.cn

Abstract:        Disk log is an important type of data structure used in databases and file systems. The synchronous nature of logging often renders it the performance bottleneck of the whole system.  One way to solve this problem is using disk-scheduling algorithms optimized for logging.  A new disk logging algorithm C latency-optimized logging - is proposed in this paper.  A High Level Petri Net (HLPN) model of a disk subsystem and subsequently HLPN models of three logging algorithms (sector-continuous logging, track-based logging and latency-optimized logging) were built and performance of the algorithms analyzed.  The results show that both of the latter two can improve logging performance by a factor of ten and the new algorithm we propose performs much better under heavy load.

Key words:    Disk logging, Petri nets, performance analysis

Supported by the National Natural Science Foundation of China under Grant No. 69873012; the National Grand Fundamental Research 973 Program of China under Grant No. G1999032702 and No. G1999032707.