Design And Implementation Of Auto-scaling Using A Hybrid Strategy Based On Cloud Platform

Cloud computing is an emerging distributed computing paradigm. It assigns computing tasks to the resource pool consisting of a large number of computers, and applications can access on demand computing power, storage space and a variety of software services. Cloud computing users can get the resources they need, and pay for usage of resources like electricity, which is convenient to access and inexpensive.More and more people begin to pay attention to and use cloud computing. One important reason is its elasticity. Cloud computing platform can make adjustments for the resources of application timely depending on the load of application. Specifically when the load of application increases, more resources is allocated to the application to ensure the quality of service (request response time, availability) and the normal operation of the application; When the load drops, idle resources is recycled to increase the utilization of resources and reduce the cost of cloud computing users.Due to the uncertainty and the relative regularity of the application load, this paper presents a solution for auto-scaling based on hybrid strategy. The solution combines neural network algorithm and static rules algorithm together and build a hybrid model of the resource requirement. It solves the problem that the prediction of the model using neural network is inaccurate.In order to verify the validity of the proposed framework, a series of experiments are designed. The experimental results show that prediction accuracy of the hybrid model has been significantly improved compared to the simple model of resource requirements based on neural network.