University of Massachusetts Amherst

Search Google Appliance


ExoGENI Federated Networked Infrastructure-as-a-Service (NIaaS) for Domain Science Applications


Friday, February 27, 2015 - 11:15am


Paul Ruth


Gunness Student Center


Recent advances in cloud technologies and on-demand network circuits have created an unprecedented opportunity to enable complex data-intensive science applications to run on dynamic, networked cloud infrastructure. This talk will discuss the ExoGENI NIaaS testbed and two current efforts to use ExoGENI technologies to support high-performance and high-throughput domain science applications.

The initial part of the talk will demonstrate ExoGENI capabilities and techniques used to support a workflow ensemble of tightly coupled ADCIRC MPI applications. ADCIRC is a widely used numerical model for computing storm surge and wind-waves. It is often used for process-based studies, risk-based studies (FEMA and US Army Corps of Engineers), and forecasting applications (North Carolina Forecast System).  

Finally, the talk will present work toward supporting dynamic NIaaS for scientific workflows. This work aims to (a) glean resource requirements from dynamic applications and workflows through proactive introspection, (b) use the requirements to construct the topology requests that dynamically instantiate application frameworks, and (c) adjust allocations dynamically based on application demands. Recent efforts have resulted in a prototype that enables the Pegasus workflow management system to monitor and predict workflow resource requirements and generate ExoGENI resource requests to fulfill these requirements.

Paul Ruth is a Senior Distributed Systems Researcher at the Renaissance Computing Institute (RENCI) at the University of North Carolina at Chapel Hill. He received his Ph.D. in Computer Science from Purdue University in 2007. His background is in virtual computation and networking for high-performance and high-throughput computing systems. He currently works on enabling ExoGENI testbed capabilities to support domain science applications.