While existing cloud services reduce the application development time, data management applications still require significant effort by cloud tenants, for often their deployment involves a number of challenges including performance monitoring, resource provisioning and workload allocation. These tasks strongly depend on application-specific performance objectives, therefore developers often rely on ad-hoc solutions for addressing them. However, such solutions eventually hinder the application's implementation and maintainability.
This project is developing declarative mechanisms that allow application developers to express custom performance criteria for data processing tasks and exploits the properties of these mechanisms to design extensible resource, workload and Service-Level-Agreement (SLA) management services for cloud databases. The project also includes the design and development of an extensible cloud-based service platform, named XCloud, which unifies these services into a usable cloud utility.
The expected results include the design of declarative interfaces for the specification of performance criteria and performance models. Subsequently, by building upon properties of these interfaces, it will result in new extensible services for resource provisioning, workload allocation as well as in the support of customizable application-based SLAs for cloud databases.
The above results will have a significant impact on the quality of cloud-based data management applications. XCloud will allow developers to declaratively incorporate existing performance modeling techniques. The combination of declarative specifications and customizable modules will allow XCloud to act as a test-bed for new performance models for cloud databases, as well as QoS-driven approaches for workload, resource and SLA management, facilitating research and innovation in this emerging domain. Finally, the research of this project will be integrated into the development of courses and research projects for both graduate and undergraduate students providing them with background on the fundamentals and practices of building web information systems and cloud data services.
SLA-driven Workload Management for Cloud Databases, D. Stamatakis, O. Papaemmanouil. In Proceedings of 6th International Workshop on Cloud Data Management (CloudDB '14), April 2014 [pdf].
Supporting Extensible Performance SLAs for Cloud Databases, O. Papaemmanouil. In Proceedings of the International Workshop on Data Management in the Cloud (DMC ’12), April 2012 [pdf].
This project is sponsored by NSF IIS-1253196, and supported by Google Cloud and Amazon Web Services through a Research Grant.