Abstract
Businesses depend on large and increasingly complex IT systems infrastructure in their operations. Such systems could be located in-house or within an external cloud. IT system resilience is a key aspect to maintaining availability and integrity of the business processes. IT system stacks include several layers and components, among others: hardware, operating systems, hypervisors, middleware, applications software, enterprise/cloud operating environments, service layers, and business process workflow managers. The resilience provided by the layers is interdependent. Each layer needs to support the overall goal of providing excellent business resilience, while managing costs. IT resilience is accomplished by adding error checking and recovery mechanisms within the components. There is a trade-off regarding which checks and recovery methods are added, and where they are introduced within the various components and layers in the IT system. For a given expectation of business resilience, the selection and location of resilience features in individual components is key to managing costs, as well as early detection of errors and quick recovery with minimal or no disruption to the business. This area of study of tradeoff analysis and optimization across the stack has also been referred to as cross-layer resilience. While the general problem above is complex and likely computationally intractable, we select and highlight some key solvable problems that need to be addressed as we make progress in this important direction. Some of the future challenges and potential solutions to developing robust business IT solutions will be presented.

BIO
Prabhakar Kudva (PhD '96) is a research staff member in the Reliability- and Power-Aware Microarchitectures Department at the IBM T. J. Watson Research Center. He currently leads system soft error rate resilience for IBM systems across the Systems and Technology Group and Research. In the past he has led the development of key VLSI CAD tools and technologies currently in use across IBM. He has received major awards from IBM and the IEEE for innovation in these areas including best paper award and innovation in research, and patent value awards. He has represented IBM at SRC and NSF, and has taught courses on the faculty of Yale and Columbia as adjunct faculty, as well as supervised several graduate dissertations at various universities. His current research interests are in cross-layer resilience of the entire stack from business process to chip technologies.