Research And Application Of Multi Tenant Pattern In Cloud Computing Environment

With the rapid development of Internet, cloud computing as a business model that breaking the traditional software sales model has been given more and more attention by internet providers. In the cloud computing environment, multi tenant model has emerged. In this model, tenants no longer need to purchase expensive servers, and also don’t worry for maintenance and upgrades of software, these problems will be handled by cloud computing providers. Cloud computing providers fee according to the usage of tenants. Therefore resources matching, and the interest relationship between tenants and providers has become a focus in the study of this thesis. The purpose of this thesis is to solve how to match the tenants and providers, and their interest relationship in the case of limited resources. For the case of profit between cloud computing providers and tenants, the main research of this thesis summarized as follows:(1) There will become a queue model when tenants request system resources. In the case of the limited system resources, cloud computing providers need to consider the resource allocation strategy to maximize profit. Tenants have different grading based on different payment. Therefore, this is a resource allocation strategy for a multi objective optimization. Existing solution is to use genetic algorithm, encode the allocation of resources on each server, and get the result after multiple iterations. Because of the diversity of the genetic algorithm, it is difficult to get the optimal result after the change of system resources dimension. This thesis presents a heuristic algorithm based on resource efficiency, it can solve the problem of diversity of genetic algorithm, and it will improve efficiency.(2) In the cloud environment, tenants have a variety need of storage resources, and also there are various types of storage resources in the cloud environment. In this case, it need to use the auction strategy of economics to form a pairing relationship between tenants and providers, and this strategy will determine the selling price of providers and the bid of tenants, thus ensure the benefits of each other. Existing solution is the auction algorithm based on grouping, grouping tenants, and matching providers. But algorithm is too idealization that the resources of providers are limited and the data of one tenant only can come from one data center. This thesis presents a general auction algorithm, it is more adapted to resources auction under common cloud environment, moreover the usage of resources of providers and the matching case of tenants will be improved.(3) Implementing a basic cloud platform due to the property of multi tenant. The cloud platform include development support environment, project management environment and running support environment. After payment and authentication, user will become tenant and can develop application, moreover he can test and deploy his application in the cloud platform.