Thesis Topic Details

Topic ID:
Declarative Configuration Management for Cloud Environments
Rajiv Ranjan
Research Area:
Cloud Computing, Services
Associated Staff
Srikumar Venugopal
Topic Details
R & D
Group Suitable:
J2EE, Database programming
Cloud computing environments are characterized by multiple providers (Amazon EC2, Microsoft Azure, RackSpace, Google App Engine, IBM BlueCloud, GoGrid, etc.) each offering software services and hardware resources under multiple, often varying configurations. In simple terms, a configuration is a set of unique metadata that make up the
"bits" and uniquely identifies Cloud-based software services (load-balancer, database, elastic IPs, etc.) or hardware resources (computing server, storage, network bandwidth, etc.). For example, Amazon's EC2 service associates following configuration metadata with a
virtualized computing server: operating system type, I/O performance band, and number of physical processors assigned, physical memory size, secondary storage size and a cost.

Further, Amazon offers virtualized compute service (EC2) under 3 categories (on-demand, reserved, spot), 9 VM types, 3 OS types, and 4 region-based pricing. Using simple
calculations, it can be computed that an application owner has approximately 324 (3 x 9 x 3 x 4) combinations to choose from. If one augments this with the problem of selecting other services (database, load-balancers) and resources (storage, network bandwidth) that may be
required for constructing and deploying application stack, then it is not hard to see the overall problem complexity. Therefore, choosing optimal software or hardware configuration for hosting applications is complicated, tedious, and error-prone, requiring significant human
involvement and domain knowledge. If this configuration selection problem is extended to multiple services and resources, it is likely to prove daunting and not cost-effective. Hence, there is an urgent need to innovate declarative configuration management database that can
help application owners in choosing the right cloud configurations based on budgets, quality
of service, service, and resource needs.

In this project, student will develop an information system under which the domain knowledge of Cloud-based service and resource configurations is formally captured by a declarative language (such as SQL). In this approach, Cloud-based services and resources are modelled as structured data that can be queried by a declarative language, and updated with well-defined transactional semantics. Through efficient and powerful rule-based reasoning on top of a database-like abstraction enables new management primitives to perform systemwide
reasoning. This prevents mis-configurations and promotes automated operations, while requiring minimum operational effort.
Past Student Reports
No Reports Available. Contact the supervisor for more information.

Check out all available reports in the CSE Thesis Report Library.

NOTE: only current CSE students can login to view and select reports to download.