Improving website scalability using mathematical model and caching scheme / Katayoon Beetarafan

Web caching has been proved to be a highly efficient method in reducing the network bandwidth consumption, as well as decreasing the network's latency. In a web cache placement strategy, an important issue is to utilize the available resources by performing a cache placement to achieve the desirable scalability. The current study has tried to explore the scalability improving strategies by taking advantage from the distributed cache placement technique. Therefore, a Finite Series Model (FSM) was introduced to analyze the hit ratio saving resulted from the copies made in the cache for the requested objects. By using this model, the scalability problem can be formulated as a function of the object requests that hit the server successfully. It was proved that the hit ratio for a single object on the origin server along adopted cache schemes follows the finite series functions. As a result, it was possible to achieve the mathematical solution by using such series. The scalability estimation and assessment was then proved by performing a case study to compare the actual data against the abstracted ones. Performance of the cache strategies were evaluated by considering the case study and taking advantage from the real workload traces. It was shown that both the client and the distributed caches performed well in the case study.