摘要：This work intends to design a new model for web cache by fragmenting the cache into three slices namely Sleep Slice (SS), Active Slice (AS) and Trash Slice (TS) and hence the name SAT for this cache model. This model is explored only for client cache. Slicing is done to group cached pages based on the hit count so that latency of retrieving them can be reduced. By discriminating one time hit pages from the remaining pages, hot pages are usually made available. Simulations have been carried out to investigate the desirable percentage of each slice with respect to the total cache capacity for different cache replacement policies. The outcome of the proposed SAT model ranked by the performance metrics such as File Hit Ratio, Speedup, Delay Saving Ratio and Number of Evictions reports thriving progress to the basic cache model. Three types of replacement policies based on the key parameters recency, frequency, size and semantic relation are considered for the model proposed. These replacement policies are implemented only in AS. SS replacement is done with respect to frequency of cached pages whereas TS replacement is done only on demand. The proposed SAT model has exhibited about 15-20% improvement on an average for the above mentioned replacement policies and performance metrics.
