With the omni-presence of Internet applications, service-oriented computing increasingly plays a dominant part in global computing. Thus, understanding how to build efficient platforms for service-oriented computing is critical in supporting continued Internet proliferation. The overriding goal of this research is to gain such understanding, and in particular to investigate basic principles in building a global Utility Computing for Internet Applications (UCIA) platform that is highly responsive in resource reallocation, provides resource isolation to different applications, and supports heterogeneous execution environments.
This research encompasses a number of specific directions. This includes studying tradeoffs between different architectures for resource sharing in a global UCIA platform and investigating efficient algorithmic approaches to request distribution and application placement that account for the computational cost of executing the algorithms and the interaction of these two policies. It focuses in particular on the interaction between two foundational technologies in global UCIA, wide-area application clustering and virtual machines. By leveraging these technologies in a novel way, this research promises a significant speed-up in resource redistribution among applications. To quantify its findings, this project utilizes a wide-area testbed running a sample of off-the shelf Internet applications and emulating several data centers.
Internet applications are transforming the Web from an information system into a global service-oriented computing platform. By improving our ability to build highly responsive, global utility computing platforms for Internet applications, this research will support continued progress in this important direction.
