SaaS Infrastructure And SLA Optimal Design Based On BDIM

With the development of society and the progress of technology, customers are not satisfied with the current method of supplying software, that is, the Software Vendors just develop software and sell them to the customers who need them and then the customers build them on their own IT infrastructure and maintain the software themselves. Using the traditional way is a waste of money and effort for the customers especially the small and medium-sized enterprises. So another method turns up. The software vendor maintains the software for customers using their own IT infrastructure and the customer pay for the software and the maintain service. This is Software as a Service (SaaS). Using SaaS we can not only reduce the economic expenditure of the customers but also can make the software be more stable, reliable by the vendors' professional maintain.With the rapid development of SaaS, the SaaS market should be standardized as soon as possible in order to ensure the rights and obligations of SaaS vendors and customers. Nowadays, there's no Service Level Agreement (SLA) standard for SaaS. SaaS vendor configure the IT infrastructure depend on experiences. This may lead to the profit be un-maximized because of not considering the technical and commercial targets. Besides, when formulating SLA they just according to experiences and the need of customers, this may be a waste of IT infrastructure and may lead to SLA violation.In this paper, we first give an advice of making SLA for SaaS and then analysis the IT infrastructure and SLA targets from a business aspect in order to find the optimal IT infrastructure and SLA targets based on BDIM which maximum the vendor's commerce profit. In order to find the optimal IT infrastructure configuration, we design a model of it. When the IT infrastructure configuration being optimized, we analysis the Service Level Objects (SLO) and build a business model of finding the property value for SLA targets by considering both the customer's requirement and the vendor's cost. In this model, we supply different SLA to different customers in different time which is more realistic. Also we have proved the correctness and high efficacy of the two models by simulation experiment.