การวิเคราะห์ผลกระทบของจำนวนแคชและวิธีการประสานงานที่มีต่อประสิทธิภาพของแคชแบบกระจาย / รชา โรจนตรีคูณ = The analysis of effects of number of caches and co-operating algorithm on the performance of the distributed web cache / Racha Rojanatreekoon
Web cache has been introduced to the internet to help decrease both bandwidth usage and user's response time. However, the increasing number of concurrent users of the internet still causes some problems to web cache. In an organization with a large number of users, a single web cache becomes a bottleneck because it cannot handle all users' requests. Thus, distributed cache has been proposed. Typically, the distributed cache has been used to share the contents among nearby caches to increase the possibility of finding requested documents and decrease document retrieval time. In recent years, distributed cache has been used to reduce the workloads of web caches and improve the overall performance. Thus, there were many studies regarding to distributed cache. Nonetheless, other important aspects such as the effect of the number of caches in distributed cache and document management schemes for distributed cache have not been researched in previous studies. In this thesis, we have studied the effect of the number of caches and cache communication method on the performance of distributed cache. In our experiments, proxy caching simulation is used as a tool to evaluate the performance of distributed cache. Using data from the office of Information Technology of Chulalongkorn University, our results indicate that the increase of the number of caches leads to the reduction of document popularity in each cache which leads to the performance degradation. However, using LRU replacement policy, which its queue is updated on every request, helps improving performance as it makes document popularity in each cache to become more realistic. In addition, document management method schemes also affect the performance of distributed cache especially the overall hit ratio and the communication overheads.