GEMA: An Object Replacement Algorithm for Cooperative Web Proxy Systems

  • Authors:

  • Ligang Dong;Bharadwaj Veeravalli

  • Affiliations:

  • Hangzhou University of Commerce, 149 Jiaogong Road, Hangzhou, Zhejiang, P.R. China, 310035. donglg@mail.hzic.edu.cn;Open Source Software Laboratory (http://opensource.nus.edu.sg), The National University of Singapore, 10, Kent Ridge Crescent, Singapore 119260. elebv@nus.edu.sg

  • Venue:
  • Multimedia Tools and Applications
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, we design and analyze a web object replacement algorithm, referred to as gain-based exchange and migration algorithm (GEMA) suitable for a cooperative World Wide Web proxy caching environment. In a cooperative environment where more than one proxy exists, the replacement algorithms used for single system cache cannot not be directly put in use to achieve an acceptable performance. In this paper, we first present an analytical model, which quantifies the “importance”, referred to as object-caching gain, of an object at a cache. This gain is used in making replacement decisions and considers the benefit of caching at local as well as neighboring proxies. Our model efficiently exploits the advantages present in the existing the research contributions on designing replacement strategies for the single-cache environment. Further, with this model, we introduce two basic powerful primitive operations, namely the object exchange and object migration, to improve an overall performance. These two operations are carried out as an outcome of replacement decisions based on the comparison of gains among objects. Thus, the calculation of the gain and deciding on which of the operations to use constitute the main part of our algorithm GEMA. For quantifying the performance of GEMA, we carry out rigorous simulation experiments based on trace-driven and event-driven approaches. Using the event-driven simulation, we comprehensively testify the performance improvement of GEMA under a variety of performance measures such as, average access time of web objects, hit ratio, and byte hit ratio. We compare and analyze our strategies with some of the popular strategies found in the literature. We also highlight some possible extensions to the research contributions in this paper.