Distributed Service Orchestration: Eventually Consistent Cloud Operation and Integration

Both researchers and industry players are facing the same obstacles when entering the big data field. Deploying and testing distributed data technologies requires a big up-front investment of both time and knowledge. Existing cloud automation solutions are not well suited for managing complex distributed data solutions. This paper proposes a distributed service orchestration architecture to better handle the complex orchestration logic needed in these cases. A novel service-engine based approach is proposed to cope with the versatility of the individual components. A hybrid integration approach bridges the gap between cloud modeling languages, automation artifacts, image-based schedulers and PaaS solutions. This approach is integrated in the distributed data experimentation platform Tengu, making it more flexible and robust.