WAP快取效能分析與影響

Abstract

無線應用協定（Wireless Application Protocol, WAP）是為了提供更有效率的無線資料存取發展而成。由於無線網路中的資料傳輸頻寬相當有限，無線應用協定便採用快取（cache）機制以縮短資料存取的時間。又因為許多無線應用服務的資料存取都具有temporal locality的特性，位於無線手持裝置中（WAP handheld devices）的快取便能利用這項特性，有效地縮短資料存取的時間。本論文探討無線應用協定中無線手持式裝置的快取效能。我們主要考慮least recent used（LRU）快取取代演算法和兩種強制一致性（strongly consistent）演算法，分別為poll-each-read和callback。在poll-each-read方法中，用戶端（client）在每一次的資料存取時，都會主動詢問伺服器（server）其位於用戶端快取中的資料是否有效。在callback方面，每當位於伺服器端的資料發生資料更新時，伺服器則會主動通知快取使該筆資料失效。在本論文中，我們提出數學分析和模擬程式的模型來探討有效快取命中率（effective hit rate）。我們的研究顯示：callback的效能比poll-each-read好。我們也在論文中探討資料存取率和資料更新分佈對於無線手持裝置中快取效能的影響。

Wireless Application Protocol (WAP) was developed to support efficient wireless data access. Since data transmission capacity of wireless link is limited, a caching model was proposed for WAP to reduce the data access time. Many WAP applications also exhibit temporal locality for data access, and the cache in a WAP handset can effectively reduce the data access time by exploiting this property. This thesis investigates the cache performance for WAP handheld devices. We consider the least recently used replacement policy and two strongly consistent data access algorithms called poll-each-read and callback. In poll-each-read, whenever a data access occurs, the client always asks the server whether the cached entry in the client is valid or not. In callback, the server always invalidates the cached entry in the client whenever a data updates occurs. We propose analytic and simulation model to investigate the effective hit ratio of data access. Our study indicates that callback may significantly outperform poll-each-read. We also report how the data access rate and the data update distribution affect the cache performance in a WAP handheld device.