Given a complex networked system of things, a configuration pertains to a multitude of system-level parameters and processes: such as algorithms, operating conditions,and network resources. Here, an optimal setting of the system configuration is critical to meet the desired application-level objectives: namely, QoS, security, and/or performance, in the face of uncertain environment conditions (e.g., attacks, faults, outages). For example, configuration control in an YouTube-like content distribution network (CDN) involves deciding on the placement of in-network content mirror sites in the underlying distribution topology for caching and rendering and the push/pull algorithmic parameters to access these mirror sites, in order to meet the content delivery latency content security, and cache-store cost constraints. Configuration decisions are continuously evaluated and replanned by an intelligent controller in the form of "observe-decide-adapt" actions, as a networked systemevolves over multiple control cycles. In this talk, I shall outline some generic optimisation methods useful for system-level configuration control based on the well-known OODA loop principles. Configuration optimisation methods and tools pertain to many things: such as the prescription of application goals and utility functions, modelling of system fault surfaces, and heuristics-aided combinatorial algorithms. I shall draw examples from the system-level case studies conducted by my group at CUNY: namely, i) cooperative control of a group of autonomous cars on the road, and ii) autonomic control of server replication for fault-tolerant cloud services. Our configuration management methods can even be applied to large-scale business process logistics and analytics.
Kaliappa Ravindran is a professor of Computer Science at the City University of New York. His research interests are in Embedded software systems, service-orientednetwork management, cloud-based distributed services, and Internet architectures. He has managed many externally sponsored research grants (both large and small) from US federal Govt. agencies and industries such as Air force, Navy, NSF, Missile Defence Agency, CISCO, IBM, General Motors, and Bell-Labs. He has about 150 research publications in well-reputed conferences and journals. He has supervised many Ph.D. students and M.S. students in the areas of distributed computing, networked systems, algorithms and protocols. His academic roots started at Indian Institute of Science 3 decades ago with B.E. and M.E degrees in Electronics and Automation disciplines. After a stint a ISRO in the area of on-board satellite control systems, he completed Ph.D. at the University of British Columbia (Canada).