An Analysis of User Behavior and Demand Swapping Policies in Time-Sharing Systems

• Item type: JInfP(1)
• 公開日: 1989-08-25
• タイトル: An Analysis of User Behavior and Demand Swapping Policies in Time-Sharing Systems
• タイトル / 言語: en
• タイトル: An Analysis of User Behavior and Demand Swapping Policies in Time-Sharing Systems
• 言語: eng
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_6501
• 資源タイプ: journal article
• 著者所属: Systems Development Laboratory Hitachi Ltd.
• 著者所属: Systems Development Laboratory Hitachi Ltd.
• 著者所属(英): en
• 著者所属(英): Systems Development Laboratory, Hitachi, Ltd.
• 著者所属(英): en
• 著者所属(英): Systems Development Laboratory, Hitachi, Ltd..
• 著者名: Yasufumi, Yoshizawa; Toshiyuki, Kinoshita
• 著者名(英): Yasufumi, Yoshizawa; Toshiyuki, Kinoshita
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: One bottleneck encountered in large scale time-sharing systems (TSS) is excessive interactive swaps. The storage management of operating systems should minimize the number of physical interactive swaps. If working sets can be left resident until users complete their inputs the number of swaps can be minimized. A demand swapping policy which maintains working sets in the main storage and swaps them only when a shortage of free storage space develops is a useful technique for resolving such bottlenecks. One important aspect of the demand swapping policy is the algorithm to determine which working set to swap out. To develop this demand swapping algorithm the trace data of the behavior of actual TSS users at a terminal are accumulated and user input processes are analyzed. Five demand swapping algorithms (LRU RAND LUFO PRED and SLRU) are proposed from the results. The number of physical interactive swaps that come from each demand swapping algorithm is compared using a trace-driven simulator. From this analysis it is found that LRU is the best algorithms among the fixed-space demand swapping algorithms. However SLRU which is a kind of variable-space demand swapping algorithm reduces the number of physical interactive swaps more than LRU within a given critical time range.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: One bottleneck encountered in large scale time-sharing systems (TSS) is excessive interactive swaps. The storage management of operating systems should minimize the number of physical interactive swaps. If working sets can be left resident until users complete their inputs, the number of swaps can be minimized. A demand swapping policy which maintains working sets in the main storage and swaps them only when a shortage of free storage space develops is a useful technique for resolving such bottlenecks. One important aspect of the demand swapping policy is the algorithm to determine which working set to swap out. To develop this demand swapping algorithm, the trace data of the behavior of actual TSS users at a terminal are accumulated and user input processes are analyzed. Five demand swapping algorithms (LRU, RAND,LUFO, PRED and SLRU) are proposed from the results. The number of physical interactive swaps that come from each demand swapping algorithm is compared using a trace-driven simulator. From this analysis, it is found that LRU is the best algorithms among the fixed-space demand swapping algorithms. However, SLRU, which is a kind of variable-space demand swapping algorithm, reduces the number of physical interactive swaps more than LRU within a given critical time range.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AA00700121
• 書誌情報: Journal of Information Processing; 巻 12, 号 2, p. 125-138, 発行日 1989-08-25
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 1882-6652
• 出版者 / 言語: ja
• 出版者: 情報処理学会