Identifying adaptation needs to avoid the vendor lock-in effect in the deployment of cloud SBAs
Proceedings of the 2nd International Workshop on Adaptive Services for the Future Internet and 6th International Workshop on Web APIs and Service Mashups
| Hi-index | 0.00 |
The migration from traditional system designs to full dynamic/elastic cloud systems raises several interesting issues, particularly surrounding the co-ordination and management of the emerging (new) system structures. Existing and emerging Service Oriented Architecture (SOA) web standards and technologies such as WSDL, BPEL and WS-BPEL are generally promoted as facilitating the design of fully adaptive and scalable enterprise, and particularly, cloud applications. However, from the end-user point of view, the functionality to provide and manage these fully dynamic cloud systems is still in its early stages and requires significant efforts to be fully achieved. Adhering strictly to software engineering concepts such as high cohesion and low coupling results in a cloud application architecture that promotes component (i.e. Service) reuse and lends itself to scalability. Equally, late runtime-binding, re-binding and fail over systems clearly highlight the flexibility of these architectures, yet their lack of adaptability is apparent when higher-level runtime alterations would be beneficial. Current efforts towards this goal, such as DADL and DURRA, still require human input at each change in order to facilitate true runtime adaptation, rendering it impractical for all but the most trivial of adaptations. This paper discusses the current cloud architecture's shortcomings along with a proposal for a new approach that is compatible with existing SOA methodologies. This approach supports XML abstraction flexibility via a new intermediary Cloud-Intention Layer, providing separation between the cloud application source code and the cloud services themselves.