The 5th IEEE/ACM International Conference on Utility and Cloud Computing
November 5-8, 2012
Chicago, Illinois, USA
Image by Matthew Rogers, used under Creative Commons Attribution-NonCommercial 2.0 Generic license

Keynote Lectures

Tuesday November 6, 2012 8:30-10:00

Clouds, Missions, Machine Learning and Reliability

Roy Campbell, Director, Assured Cloud Computing Center, University of Illinois at Urbana-Champaign

Clouds are expanding into diverse roles: big data with data mining and machine learning; real-time data and analytics with streaming (Storm Clouds); and massive andcomplex cyber physical systems with machine learning and decision algorithms. Economic pressure and the global reach and extensive future deployment of mission critical applications on clouds provide a new set of challenges for research. The reliability and security of clouds and their applications including big data and machine learning will have a major impact on real-world phenomena and mission safety. However, the current cloud solutions are orders of magnitude less dependable than the minimum requirements for mission critical systems including existing commercial and cyberphysical systems. This talk will review research progress in the Illinois Assured Cloud Computing Center.

Wednesday November 7, 2012 8:30-10:00

Predicting the Unpredictable in Complex Information Systems

Kevin Mill, Senior Research Scientist, Information Technology Laboratory, US National Institute of Standards and Technology (NIST)

Modern society exhibits an inexorable trend toward reliance on large, distributed information systems built on the global Internet, cloud computing and mobile devices. Our ability to understand, predict and control such complex information systems lags far behind our engineering and deployment skills. Over the long term, this growing knowledge gap threatens society with significant disruptions, arising from low-probability scenarios that can incur substantial costs. In this keynote, Dr. Mills outlines previous NIST research that investigated and evaluated techniques to understand and predict macroscopic behavior and user experience in complex information systems, such as computer networks and infrastructure clouds. Dr. Mills also describes ongoing NIST research seeking (1) design-time methods that enable system architects to identify and evaluate global failure scenarios that could lead to system collapse and (2) run-time methods that alert system operators about incipient transition to global failure regimes. By sharing this information, Dr. Mills hopes to inspire further research into methods to improve our scientific understanding of the complex information systems on which society increasingly depends.


Thursday November 8, 2012 8:30-10:00

Elastic Software for Cloud Computing

Carlos A. Varela, Worldwide Computing Laboratory, Rensselaer Polytechnic Institute

Flexibility of software is critical to meet the demands of cloud computing given the users’ expectation for on‐demand scalability. For example, software flexibility must enable to execute a large‐scale scientific simulation partly on a private grid computing resource, and partly on a public cloud or volunteer computing cloud, according to specified declarative user policies, which may favor cost, time, or energy. Software flexibility must be supported at multiple levels of the software life cycle: at the programming layer, at the compilation layer, and at run‐time through middleware for adaptive distributed systems execution.

In this tutorial, we show how to use the actor model of concurrent computation to develop elastic software for cloud computing. In particular, we present SALSA, an actor programming language that enables developing dynamically reconfigurable open distributed systems. Then, we introduce IOS, a middleware layer that uses work-stealing through actor migration to accomplish load balancing of distributed SALSA programs. Finally, we briefly discuss COS, another middleware layer to provide elasticity to SALSA programs, so that applications can scale up and down within private clouds, as well as scale out to IaaS public clouds.

© IEEE Technical Committee on Scalable Computing • 2012