M.Sc. Tezi Görüntüleme

      Web caches are used as a convenient solution for network load reduction and user access speed improvement. With the rise of modern technologies like VPN to secure connections, placement of web cache servers in optimal locations has become a serious concern in WANs and ISPs. Web cache replacement policies are the main part of web cache servers. In case the memory is full, these algorithms replace old or infrequently used objects with those that have higher probability of usage by new or future users. This article consists of two studies. First study proposes a dynamic web cache replacement algorithm which trains itself utilizing previous network logs and by exploiting the data mining clustering algorithm. It shows the best performance on different networks by scoring each metric of the enquiries including recency, frequency, size and delay. Considering the connection load, bandwidth and traffic characteristics, second study proposes a method for optimal placement of web cache servers. According to simulation results, proposed algorithm has better performance compared to other traditional algorithms in different network according to HR, BHR and DR. The simulation results of cache placement show that the web caches with higher priorities save more bandwidth compared to caches in next queues. By graphing the normal logarithms of the saved bandwidth ratio and priority numbers it has been shown that the prioritized caches comply the Zipf’s law.