In this talk I will give an overview of work on large industrial and eResearch systems in our software engineering research group. These systems are typically cyberphysical and cloud-enabled. This means software-intensive networked physical infrastructure with embedded software for monitoring and control is connected to IT systems in data centres to achieve real-time, distributed and parallel processing and analytics beside high productivity. We look at these systems from a services configuration and composition point of view. In our industry-funded and eResearch grants my group is trying to grapple with performance, reliability and other extra-functional properties that we have shown to be highly architecture-dependent. eResearch has become a synonym not only for high-end computational methods underpinning global computing cluster and grid services for physics, molecular biology or life sciences. Increasingly the engineering sciences look at such systems from the viewpoint of collaborative engineering support in automation systems. For example in research contracts with ABB we have applied this to collaboration support services around robotics and plant control and with Siemens to improving cloud-based testing of engineered systems. Moreover, more and more also arts, humanities and social science are transforming into ‘digital humanities’ requiring computational analytics, search, tagging and digital curation services, partly connecting researchers with citizen scientist crowds. Accordingly, such services are becoming extreme. They are characterised by massive data volumes and speeds, high-performance computing methods, same-time collaboration and visualisation platforms, laboratory instruments that are remote controlled. And increasingly these services are based on real-time connectivity to infrastructures in plants or laboratories at different global locations. This talk provides a snapshot of eResearch services in the Australian context with a focus on our own work at RMIT University. We touch on a few specific challenges in architecture variation, dependencies and performance/reliability tradeoffs that are highly dependent on the systems architecture.
Heinz Schmidt is Professor of Software Engineering in the RMIT Computer Science School. He is Director of eResearch, a university wide role to help accelerate research through advanced use of ICT. He also directs the Australia-India Center of Automation Software Engineering (AICAUSE). Heinz received his PhD from Bremen University, Germany. Heinz is also Adjunct Professor at Mälardalen University, in Västerås (near Stockholm) in Sweden, where he works with the Mälardalen Real-Time Research Centre and the Embedded Systems Group in their CS department. Prof Schmidt is internationally recognised in software engineering and software architecture for parallel and distributed systems. He has over 30 years experience with object-oriented and component-based software architectures, languages, systems and tools, in practice, research and training. Before joining RMIT, Heinz held positions at Monash University, the CSIRO, ANU, the International Computer Science Institute of UC Berkeley and the German Research Centre for Mathematics and CS (now transitioned into Fraunhofer). Over the last few years his research has particularly focussed on modelling, prediction, verification and testing of extra-functional properties in software-intensive distributed platforms. Such properties include availability, reliability, performance and safety. Heinz has led industry collaborations in Australian CRC contexts and European projects.