OW2con'19

Presentation of the open source MELODIC platform developed within H2020 MELODIC project. The platform allows for mulitcloud, cloud provider agnostic deployment of the application with advanced optimization capabilities. MELODIC is integrated with AWS, Azure, GCP and Open Stack based cloud
providers and supports Virtual Machines, Containers, Big Data Frameworks (Spark) and serverless components. Due to highly advanced optimization capabilities using both deterministic (Constraint Programming Solver) and reinforcement learning methods (Learning Automata Solver) MELODIC is able to solve complex optimization problems. Using sophisticated utility based approach is possible to find the best deployment and automaticaly deploy all components. Thanks to built in mulit layer metrics gathering system based on ESPER MELODIC is able to monitor applications and theirs metrics. In case of the violation of the metric value MELODIC automatically reconfigures applications looking for better deployment configuration. The above capabilities sets new standard for multicloud optimization and deployment. Comparing to container based schedulers like Kuberenets MELODIC allows for multi type components deployment (Vms, Big Data, containers, serverless) and dynamically adding and removing cloud resources, based on the current needs of application.