Service composition pattern generation for cloud migration: A graph similarity analysis approach

Abstract

The demands of migrating existing on-premises complex enterprise applications to cloud dramatically increase with the wide adoption of cloud computing. A recent research validates the possibility to combine multiple proprietary migration services offered by different vendors together to complete cloud migration. Pattern based service composition has been proven as an appealing approach to accelerate the service composition and ensure the qualities in the Service Oriented Architecture (SOA) domain and can be applied to the cloud migration service composition theoretically. However, current pattern generation approaches are not applicable for the cloud migration due to lack of either existing cloud migration business process knowledge or execution logs. This paper proposes a novel approach to generate cloud migration patterns from a set of service composition solutions. We formalize the pattern generation as a special graph similarity matching problem and present an algorithm to calculate the similarity of these service composition solutions. Patterns are chosen out of the solutions by similarity with designed criteria. The benchmark results and quantitative analysis show that our proposed approach is effective and efficient in pattern generation for cloud migration.