An adaptive and cost-efficient migration to cloud approach in dynamic environments

Due to the variety in available cloud providers along with the frequent changes in strategic objectives of an enterprise, migrating existing software components to the cloud has become a challenging decision in the software maintenance phase. "Financial" and "customer satisfaction" viewpoints are two important strategic objectives of all enterprises that greatly affect the decision about migration to the cloud. Moreover immense number of target cloud services with too many configurations and cost models has made the search space of possible migrations huge. Many existing approaches of software migration to the cloud have modeled the problem as deployment optimization of software components over available platforms, while in this research following a valid migration plan is intended rather than proposing a final optimal migration solution (deployment). A migration plan is a sequence of actions to be taken by the technical team to move the software components to the cloud stepwise. Since at each stage of the migration there might be many valid alternative paths to follow, a recommender module is proposed to direct the management by recommending the best migration plan out of all valid plans in a Labeled Transition System. The recommendation is based on the current state of the enterprise which is estimated using a two-state Hidden Markovian Model by observing ambient signals. The empirical study showed that particularly in dynamic and changing conditions the proposed adaptive and plan-oriented method succeeded in posing lower accrued maintenance costs on the enterprise over time with confidence 90% compared to the non-adaptive method due to its reactive and self-balancing nature.