A Model Of Dynamic Computing Resource Management Based On Multi-Tier Application On Clouds

As a new computing model, the advantage and success of cloud computing are promoting many service providers and enterprises to migrate their existing applications into clouds in an economical manner. Dynamically-scalable and managed computing resource, such as storage, platforms, and services etc., is delivered as"pay-per-use"manner to external customers over the Internet via this new computational paradigm, which greatly improves the performance and decreases the cost of Web applications. Resource management is one of the most essensial problems in Cloud Computing. Many existing researches have been proposed on solving resource management problem for the purpose of maximizing the utilization of virtual resource. However, rare researches have been proposed related to the problem of how to maximize the profits of Web service providers.In order to achieve high profits, cloud-based web application providers should carefully balance cloud resources and dynamic workloads. However, this task is usually difficult because of the complex nature of most web applications. Furthermore, service providers can also optimize global resources with minimal quantity and maximal utilization, in order to maximize their profits. First, performance model should be effectively adjustable to workloads that fluctuate over time in order to support next dynamic scheduling. Second, provisioning mechanism have to detect which tier is responsible for the changing of the number of VMs. Third, scheduling decision will be made according to service provider?s interest to maximize revenue generated via providing Web application. It is also profitable and efficient for the sake of cloud providers, because the utilization of virtual resource has been greatly improved.In this paper, we presented a predictive performance model to analyze such applications and to determine when and how much resource to be allocated to each tier of an application. Our model is more suitable for cloud environment for the reason: (1) it supports multiple services running in heterogeneous environment; (2) it?s a self-adaptive open model for changing environment. In addition, we proposed a new profit model to describe revenues specified by the Service Level Agreement (SLA) and costs generated by leasing resources. It is in the service providers ?interests to minimize the cost of using resource provided by cloud infrastructure vendors and maximize the revenue (specifically, service profits) generated through offering services. Our proposed mechanism solves the problem how many virtual machines (VMs) a service providers should request from cloud (e.g. Amazon, Azure) in response to the observed services? performance and profits? fluctuations.We also designed and implemented three experiments to validate performance model and profit-driven scheduling policy, including workload experiment on MATLAB, simulation experiment on CloudSim and RUBiS based experiment that adopted our proposed methods. Experimental results indicated that our model faithfully captures the performance and resources are allocated properly in response to the changing workload, thus the goal of maximizing the profit has been finally achieved.In this paper, we addressed the new challenges raised by profit-driven provisioning of multi-tier Web application in Clouds. We proposed a dynamic profit-driven scheduling technique for multi-tier application that employs (i) a performance model to analyze and predict such applications and provide insight into application?s behavior to determine how much resources to be allocated to each tier of the applications, enabling service providers to allocate their computing resource with flexibility, and (ii) a predictive and reactive method that determines when to remove or add these resources, and (iii) a profit-driven provisioning technique to maximize profits of SaaS provider. This work showed promising application of Cloud computing as this paradigm is primarily driven by its cost effectiveness. As the scheduling of composite multi-service Web application, particularly in cloud system, has been intensively studied, our proposed technique allows multi-tier web applications to overcome their performance limitations, non-scalability and poor availability by taking advantages of Cloud computing infrastructures.