The engineering of multi-Cloud systems is a less explored field and poses a series of challenges due to the current weaknesses of the Cloud computing practice:
W1. The lack of common API for provisioning and accessing resources from multiple clouds.
W2. The lack of engineering techniques and tools for multi-Cloud applications.
W3. Existing autonomic architectures and techniques not adapted for multi-Cloud environments.
The first weakness and corresponding challenge to overcome is addressed by the FP7 mOSAIC project (Sept 2010- Feb 2013), scientifically coordinated by this project team. mOSAIC is building a PaaS solution (second version available in early 2012) that will provide portability of the Cloud compliant applications, based on a Cloud ontology, semantic processing, negotiations using multi-agent systems and a simple resource provisioning mechanism.
The new project, AMICAS, will take advantage of the mOSAIC API and platform to develop its solutions to cover the other two weaknesses and the corresponding challenges:
C1. Ensure trade-off between increasing reliability by scaling components over multiple Clouds and reducing costs.
C2. Build techniques and tools for automated multi-Cloud environments capable to be scaled and distributed across several Clouds and to take management decisions.
In order to describe the proposed approaches for dealing with the first two scientific objectives, we enumerate several artifacts that will represent the outcome of the project:
- Architectural styles for scalable and robust multi-Cloud applications.
- An architecture for autonomic management of multi-Cloud applications.
- A technique for making a trade-off between reliability and cost when making the deployment of multi-Cloud applications.
- Techniques for dynamic load balancing in Sky computing environments.
- Algorithms for scheduling in Sky computing
- Adaptation of reactive and proactive self-management techniques to the context of Cloud and Sky computing.