Distributed applications comprising components from multiple administrative domains, pose scalability and availability challenges because of the lack of a central administrative authority. This project investigates, in the context of the Web Services architecture, whether application-level intelligence on the network path between clients and services can help address these challenges. The intelligence is embodied in modules for inspecting client traffic, making decisions about service actions such as replication, request redirection, or admission control, and finally realizing these actions. The project combines a system architecture for executing service-specific actions at intermediate routing nodes with application-neutral algorithms for servicing client requests closer to the network edge. Both the architecture and the algorithms work with a semantic framework, which treats service requests as structured accesses against a (physical or virtual) database.
The intellectual merits of the activity include (1) development of novel techniques and algorithms for automatically scaling component-based services in response to adverse client and network load conditions; and (2) investigation of architectures for securely and efficiently realizing service-specific actions on client requests closer to the network edge.
Broader impacts of the activity include (1) techniques for improving the resilience of the nation's cyber-infrastructure against malicious attacks; and (2) integration of the research results into an undergraduate-level course that will train students to build component-based distributed systems.
