Most vendors of e-commerce applications deploy the cache memory to deliver the web objects to clients faster. However, they face many problems in dealing with the cache memory due to limited resources and dynamic access patterns. Especially, the frequently changed demands of customers and insufficient memory for thumbnails of products in online shopping mall make the performance low. As a result, we need to efficiently manage the cache memory by evicting the unused data. The performance of cache manager depends upon the efficiency of delete determination. In this paper, we propose reference table based cache eviction strategy using ERF, which is an extension to the conventional LRU and LFU algorithms by considering the time when the Web objects are referenced and the frequency of references while the items present in cache as main factors for the cache replacement. ERF uses natural exponential function on time to cope with dynamic nature of e-commerce business with limited memory. It sorts the caches in the order of score value which come from coordination between frequency and recency and evicts the caches according to it. According to a predefined score function, we can keep the most important contents in the cache. We evaluate the performance of reference table based approach in terms of CPU consumption. We also evaluate the performance of RERF by using the workload which reflects the real-world applications and compare it with conventional algorithms. By increasing the cache hit ratio with RERF, we can expect the decrease of copy and delete operations of cache with improving the overall system performance.
