Joint VM and container consolidation with auto-encoder based contribution extraction of decision criteria in Edge-Cloud environment

In the recent years, emergence huge Edge-Cloud environments faces great challenges like the ever-increasing energy demand, the extensive Internet of Things (IoT) devices adaptation, and the goals of efficiency and reliability. Containers has become increasingly popular to encapsulate various services and container migration among Edge-Cloud nodes may enable new use cases in various IoT domains. In this study, an efficient joint VM and container consolidation solution is proposed for Edge-Cloud environment. The proposed method uses the Auto-Encoder (AE) and TOPSIS modules for two stages of consolidation subproblems, namely, Joint VM and Container Multi-criteria Migration Decision (AE-TOPSIS-JVCMMD) and Edge-Cloud Power SLA Aware (AE-TOPSIS-ECPSA) for VM placement. The module extracts the contribution of different criteria and computes the scores of all the alternatives. Combining the non-linear contribution learning ability of the AE algorithm and the intelligent ranking of the TOPSIS algorithm, the proposed method successfully avoids the bias of conventional multi-criteria approaches toward alternatives that have good evaluations in two or more dependent criteria. The simulations conducted using the Cloudsim simulator confirm the effectiveness of the proposed policies, demonstrating to 41.5%, 30.13%, 12.9%, 10.3%, 58.2% and 56.1% reductions in energy consumption, SLA violation, response time, running cost, number of VM migrations, and number of container migrations, respectively in comparison with state of the arts.