Research On Multi-tenant SaaS Service Customization And Deployment Method Based On MapReduce Ant Colony Optimization

The rise of the cloud computing is gradually changing the entire computer industry and academia. Cloud computer reflects the idea of "network is computer". It forms huge shared virtual IT resource pool, which combines a large amount of computing resource, storage resource and software resource, to provide the user of the remote computer with IT service of infinite ability, which comes as soon as called. Cloud computing is divided into three level, including Infrastructure as a Service (IaaS), Platform as a Service (PaaS) and Software as a Service (SaaS)SaaS turns the ownership of software, running of the software and infrastructure, management and maintenance to external operators. The user acquires the right to use by the means of paying through the Internet, rather than having the software or added hardware. The delivery mode of SaaS provides the user with the serving pattern of applied software. The user is able to reduce the cost of structure, use and maintenance of the applied software and increase the flexibility of the change of the business, through renting the software.Data center is the fundamental of cloud computing. With the extension of the scale of data center, the maximum cost of running and maintenance of data center is the energy consumption. The emerging problem of energy consumption is a serious impediment to the popularization and development of cloud computing technology. Many manufactures is now actively doing the research of green energy saving technology, expecting to seize the maximization of self-interest through quickly grabbing the commanding heights in the field of energy saving.The programming mode, distributed storage of huge mounts of data, managing technology of huge mounts of data, virtual technology and the managing technology of cloud computing platform are included in the key technology of cloud computing. The cloud computing technology solves the problem of the storage and rapid processing of huge mounts of data. It makes millions of cheap computers coordinates with each other to finish the mission of cloud computing together. There is a natural connection between the cloud computing and intelligent algorithm. The MapReduce programming mode of cloud computing is originated from the Lisp language of artificial intelligence and swarm intelligence algorithm, such as Ant Colony Optimization (ACO), Genetic Algorithm (GA), Simulated Annealing (SA) and algorithms like that. The swarm intelligence algorithm is well applied in the cloud computing environment because it uses the method of Monoto Carlo in large quantities, which has high parallelism and can realize distributed parallel computing, and is able to play a cloud computing system strong computing power and storage capacity.Ant Colony Optimization (ACO) is a kind of bionics group intelligent algorithm optimization. It is inspired by the scientist’s research of ants foraging model. It has self-organization ability, positive feedback ability, strong commonality, robustness and high implicit parallelism.In this dissertation, ant colony algorithm under cloud computing environment and multi-tenant SaaS platform service customization and energy perception of service problems is studied. The following are the content and innovation points of the research.1. The research of ant colony algorithm under cloud computing environment. Several key technologies of cloud computing and ant colony algorithm are combined and distributed and parallel ant colony algorithm under cloud computing environment is designed. MapReduce-based Improved Ant Colony Optimization for Multidimensional Knapsack Problem (MIAM)[104] is put forward. The general train of thought, method, characteristics, structure and performance of solving the problem of this algorithm which give full play to the strong capacity of computing and storage of cloud computing system is studied to provide with new train of thoughts and methods of solving the problems intelligently, distributed and in parallel and solving the intelligent management problems of cloud computing. The programming mode of MapReduce which realizes the parallel computing of ant colony optimization algorithm and the methods of roulette, intersection and variation is applied to improve the ant colony algorithm. Computational complexity of ant colony algorithm is decreased through changing the probability calculation time. Futhermore, the distributed and parallel solving of multi-dimension and large scale problems of this algorithm is applied.2. The ant colony algorithm under cloud computing environment is applied to solve multi-tenant services custom problem in SaaS platform. Multi-tenant services custom can meet the changing personalized service demand of the tenant and is also one of the key technologies of flexibly realizing the SaaS multi-tenant structure. The related theory, key technology and implementation method in SaaS is studied. The hierarchical structure and order process of multi-tenant services is put forward to improve the quality and efficiency of service of SaaS platform with theoretical significance and practical application value. A customization algorithm based on MapReduce and multi-objective ant colony optimization (MSCMA) has been proposed. MSCMA customize the most suitable for the business process and optimize the service portfolio for the tenant from a number of business processes and mass services. Multi-objective ant colony algorithm is designed and the MapReduce cloud computing technology is applied in the MSCMA. This dissertation studies distributed parallel operation optimization task in a cloud computing environment and apply the good solution keeping strategy and diversity strategy. The outcome of the simulated experiment shows that MSCMA performed favorable convergence and scalability in solving multi-tenant service customization; it also demonstrates the proposed algorithm has good ability in processing massive data and large scale problems.3. The ant colony algorithm under cloud computing environment is applied to Solve the problem of energy perception of service place. Service group set and storage strategy and algorithm in SaaS platform is designed to generate idle servers and user service requests can be distributed to the data center right amount on the server. The operating costs of data center is decreased by turning off unused servers to reduce energy consumption, which has important application value and conforms to the development of low-carbon economy and green computing concept and the overall development trend of cloud computing. The service deployment algorithm is designed and the service deployment algorithm based on MapReduce and multi-objective ant colony algorithm is put forward. This algorithm solves mass service deployment issues distributed and in parallel and meanwhile attain the objective of minimum cost, the deployment server load balancing, maximizing meet user requirements, etc. In addition, this algorithm is suitable for various scenarios of service deployment issues.