The Service Composition Research Of Cloud Computing Based On The Auction

As an emerging computing model, cloud computing is able to provide online services to the consumers. In the cloud, users can lease resources on-demand from the cloud service providers, without having to purchase and maintain expensive hardware and software resources, thus reducing the operating costs. Currently, the services provided by CSPs are relatively simple. In many situations, they cannot meet the demands of increasingly complex needs of users. To solve the above-mentioned problem, a feasible solution is to build a CSP dynamic cooperation platform. Through this platform, multiple CSPs collaborate together to meet the complex requirements of users.The DCP (Dynamic cooperative platform) can help reduce cost of users and enhance the capacity of CSP which provide users with a combination of cloud services. However, there are there significant challenges involved in creating such a platform:First, how to find a suitable market model to ensure that the platform running normally and efficiently; Second, how to design the cooperation platform framework in order to minimize conflict between CSP; Third, a CSP how to choose partners, this is clearly a NP-hard problem, so we need to find a approximation algorithm to ensure that a CSP can find the right partners in polynomial complexity. In this paper, we proposed a coalition-auction-based market model, built a CSP-DCP, designed the coalition formation algorithm, specific details go as follows:1. Built A CSP-dynamic cooperative platformThe platform adopts the coalition-auction market model in order to reduce platform service prices, and the framework of the platform is divided into three levels, the first layer is user layer, including all users that submit service request to the agent; The second layer is auction layer, including the following two functional modules:①information release module, this module is responsible for accepting user requests and to release it to the market;②the auction module, receive all coalition price, and use the second price auction to choice the won coalition. The third layer called the CSP decision-makers layer, this layer include the following modules:①the task search module, CSP selects the appropriate user service requests according to their own resources in the market;②partner selection module, when one CSP cannot satisfy the user request alone, it will find collaborators through the coalition formation algorithm;③negotiation module, the main content of the consultation around the coalition member price. Negotiation process is divided into two steps, each step in the coalition members will be under the personal data to negotiate between the two.④payment module:According to the optimal and sub-prime coalition members price, we can confirm the price that optimal union paid to members finally, and then submit to the agency coalition price.⑤evaluating module:The sponsor of coalition evaluates the formed coalition using the historical transaction data. The pCSP will bargain with the partners if the quotation is too high.2. Designed two kinds of partners selection algorithmAccording to the different form of the services, we put forward two kinds of CSP partner selection algorithm to solving the problem of coalition formation. One is the PSA based on the biding on the single services, including the N-based PSA and the AC-based PSA. Another one is the PSA based on the biding on the composition services.3. Provided a prototype system of DCP.According to the design of cloud platform framework, using the MyEclipse to simulate the entire cloud platform of collaborative process, we have realized the function of each module, and using MSSQL to store data produced in the platform operation process, using the XML records service demand template.