Optimal Design For Infrastructure In Multi-Layer Service System

In the highly complex distributed computing environment, network application services systems tend to be refined for many services subsystem to form the multi-level structure. This structure can easily expand and is very versatile, which allows to pass messages on different machines across the network interoperate. In the use of this service, users only care about the service suppliers to provide the quality of service and effectiveness of services in the face of a whole object name service interface, which do not care about to establish what language running on what operating systems and hardware platforms. So how to design a optimal infrastructure for service providers to meet customer service needs is particularly important. Good infrastructure design can help service providers conserve resources costs, under the premise of ensuring the performance of services to reduce commercial customers pay for the service costs, and improve customer service satisfaction. As a result of this multi-level parallel service system has the advantages of reusability, independence, if the customer needs to change, or a layer of service changes to upgrade the service provider does not need to change the entire service, just to be changed to the appropriate level of service, so design a good infrastructure for the entire service system is also essential.There have been some studies have looked at the service infrastructure design, but most of the research is to stay in a single service system, or a given the system model and cannot solve the problem of complex system simulation exist today, such as clouds computing service. Therefore, this study proposed model is based on the reality of service scene, parallel service request multiple service requests in parallel. While the model can be processed through the network service request issued by the customer at any time or anywhere, and the request is not a single type of service requirements, may be a mixed type of service request.In order to better improve customer service satisfaction, Firstly, this study is research on how to design an appropriate SLA, so that the customer and the service provider can agree on the issue of service performance which discussed in the SLO indicators to find the most suitable for the service, and then find a design approach makes the SLA decomposition, dispatching the top-level SLO to low-level SLO, resulting in each underlying service performance thresholds, and ultimately get the optimal structure design of infrastructure. The greatest degree of design obtained by this method can meet the expectations of the customer, and considering both the performance of IT metrics and the commercial interests, to ensure the economic benefits of the service.Finally, we used a cloud-based business website to verify the approach. The validation scenario is a large number of connections through the network computing resources unified management and scheduling, to form a pool of computing resources to provide on-demand service to users. Cloud computing services is opaque to the user, the user would look at it as a whole, users concern just do not care about cloud computing service provider to provide the service and its quality of service, how to implement the service, and how to deploy the service resources. In the verification process, the research will be based in the cloud computing environment, the SLO target value given by the user and the service provider, the proposed method give a reasonable infrastructure structural design, at the same time, we test other methods to get the infrastructure comparison with our approach, and to prove the effectiveness of the proposed method.