Optimizing the access performance and data freshness of distributed cache objects considering user access pattern

Caching has long been used in most fields of the computer systems to enhance the scalability of the objects, improve the performance and reduce the access latency. A significant effort has been made to introduce cache-coherent algorithms for maintaining the consistency of such data objects in cache by keeping a higher freshness of the data. Updating the cache objects considering the access behavior and user preferences is one of an attractive solutions to maintain the consistency. In this paper, we define quality of data (QoD) metric to evaluate the amount of freshness that is necessary to satisfy the user requirements. We then focus on the update scheduling method that analyzes the access behavior of the cache objects and predicts the time interval for updating the cache. Here, we introduce the "average update interval method" that uses the most recent time between access values, to predict the time interval. Using our proposed algorithm, the user can not only access the preference view but also he can get the maximum QoD of the objects. Moreover we performed extensive experiments using web log data and simulation data. Then the results could conclude that the cache objects are maintaining more than 70% of consistency with the original objects.