Distributed and Networked Systems

Distributed and Networked Systems

Modern systems such as factory floors or aircrafts and cars consist of many components that are specialized for certain tasks. Nonetheless, the overall system assembled from these components needs to function smoothly. For example, modern cars possess a multitude of electronic control units (ECUs) to provide the vehicle’s overall functionality. These ECUs work simultaneously and need to be interconnected in order to exchange data such as measurements and commands.

The boundaries of a distributed system are often less obvious and within short distance, e.g., when functionality depends on access to remote data or compute power over the mobile network and the Internet. Independent of size or the underlying network technology, such a collection of independent computers is called a distributed system as long as the user experiences it like a single, coherent system.

Challenges

Challenges in the design of networks and of distributed systems are manyfold. They range from functional aspects, i.e., realizing the desired functionality, to non-functional aspects like performance, security, scalability or (energy-)efficiency of the proposed system design.

In the future, the use of applications that allow no tolerances in terms of fast, reliable and predictable data transport, for example remote cardiac surgery, will increase. Thus, research touches many areas of modern computing systems. Among them are the design and analysis of hardware and software as well as their integration, simulation and measurement of system behavior, as well as analysis and optimization of algorithms with respect to non-functional aspects.

Focus areas

The research area Distributed and Networked Systems with Prof. Dr. Steffen Bondorf is mainly concerned with the following topics:

performance modeling: understanding and manipulating complex dependencies within systems in order to infer and tune their performance. This is pursued in the context of distributed and networked systems, mainly in their interpretation as systems of interdependent queues.

deterministic performance analysis: analysis of deterministic networks using the Network Calculus.