Performance Estimation for Business Workflows on Public Cloud Offerings Using Probabilistic Model Checker

Because of the prevalence of public SaaS cloud offerings, such as e-mail and spreadsheet (document) services, many cloud users are beginning to construct their cloud business processes by cascading various cloud offerings as workflows that are executable on workflow engines, such as Fujitsu´s Run My Process. Although utilizing existing cloud services when constructing business processes can reduce the time and costs required for development, it is difficult to estimate the availability and costs required for executing workflows with complex structures during the design stage. To solve this problem, we have developed a method to estimate the quantitative aspects of business workflows when using public SaaS cloud offerings. Using the probabilistic Markov decision process (MDP), we have constructed models that represent the behavior of workflow utilizing cloud offerings. Then, using the PRISM probabilistic model checker, we have calculated the models´ performances. A case study shows that our approach can estimate the performance of a complex business workflow before it is executed.