Autonomic Applications Research
The overarching goal of this research is to develop novel concepts and tools that will enable users to develop a new generation of ground-breaking scientific and engineering simulations on the Grid. Experience has shown that the phenomena being modeled are inherently complex, multi-phased, multi-scale, dynamic and heterogeneous in their temporal and spatial computational demands. Likewise, the Grid offers a dynamically patterned computational infrastructure, globally networking large numbers of idiosyncratic computing and communication resources, data stores, and sensor networks. The combination of these two characteristics presents grand challenges to the computational and computer science fields. Our underlying hypothesis is that these challenges can be overcome by coupling recently developed concepts of autonomic computing with new modeling approaches that exploit and match the heterogeneity of the applications to that of the Grid. We propose to develop an innovative programming paradigm and a supportive Grid middleware infrastructure that are based on autonomic computing strategies. These solutions will allow scientists to program on a Virtual Grid (vGrid) that will automatically compose, map, adapt, optimize and manage the execution of next generation science and engineering simulations. The vGrid concept will lead to fundamental innovations in the development, optimization, deployment and management of applications and resources on the Grid.
Autonomic runtime management of SAMR applications aims to define, design, implement and deploy the autonomic application components in the AutoMate framework. Its overall objective is to enable the adaptive and agile self-management of autonomic components to orchestrate the runtime management of SAMR applications in order to achieve the efficient utilization of resources and the optimal performance.
The Grid community is actively working on defining, deploying and standardizing protocols, mechanisms, and infrastructure to support decentralized, seamless, and secure interactions across distributed resources. Such an infrastructure will enable a new generation of autonomic applications where the application components, Grid services, resources and data interact as peers. In this effort we describe the development and operation of a prototype application that uses such peer-to-peer interactions between services on the Grid to enable the autonomic optimization of an oil reservoir.