We envision a new architecture of databases that dramatically reduces administration costs while managing a large scale data reliably. The key is to provide data management capability as a service. This vision, called CloudDB, is along with the current trend of "Cloud Computing" that allows users to access IT-related services in the cloud without the knowledge of the underlying technology infrastructure. Cloud Computing is a fundamentally different computing model and traditional data management systems are not designed to work in this new environment. Consequently, data management in cloud computing raises particularly challenging problems, which call for pioneering technology development and innovative solutions.

CloudDB is a comprehensive data management platform in the cloud. The envisioned system would provide data management capabilities as a service to transparently and efficiently support heterogeneous application workloads with identifiable SLA guarantees and end-to-end system management functions. The system will be able to employ heterogonous underlying storage models to effectively meet applications' query and scalability requirements. We propose the achievement of data independence for all underlying specific storage models as the key guiding principle for the system. If the system is able to achieve data independence, the application logic is decoupled from the data processing logic and allows applications to enjoy the benefits of individual storage models, which are optimized for particular purposes, without worrying about the specifics of data processing.