Enabling Scalable Cloud Computing

Event Sponsor: 
Mathematics and Computer Science Division
Start Date: 
Apr 9 2015 - 11:00am
Building 240/Room 4301
Argonne National Laboratory
Ketan Maheshwari
Speaker(s) Title: 
Postdoctoral Researcher - MCS
Dan Olson
Scientific computation landscape is a complex ecosystem comprising of many elements (eg. data, computation, parallelism), components (eg. software, hardware, application), and players (eg. researchers, tech-leads, system
Leveraging the capacity and capability of such an environment is a significant challenge. For instance, porting applications on large scale computing systems is a non-trivial problem. A cloud environment can benefit the application space while at the same time helping admins managing the hardware resources such as storage and network. Cloud frameworks can abstract data management and computation for users. Computation gets available in compact, self-contained boxes while access to data is abstracted as accessible objects.
HPC (eg., supercomputers, clusters), and HTC (eg.  grids) infrastructures pose operational challenges such as access methods, policies, network restrictions, resource managers, and data movement. Finer details are distinct for each system and consequently the approaches to address them differ. A cloud computing layer on top of such infrastructures can unify and simplify access by the users community.
This talk will discuss the strategies my research employs with the aim of seamless and scalable application execution from initial porting to final results dissemination. 
Ketan Maheshwari is a postdoctoral researcher in the Mathematics and Computer Science Division at Argonne National Laboratory. He earned his doctoral degree in 2011 from the University of Nice with highest honors. Maheshwari's research interests includes distributed and parallel computing with an emphasis on workflow programming of high-level science and engineering applications. Maheshwari's recent paper on clouds, co-authored with colleagues and collaborators has won the best paper award at the 2014 HPDC workshops.