Persons involved: Martin Fränzle, Jorge Marx Gómez, Wolfgang Nebel, Achim Rettberg, Oliver Theel, Andreas Winter

As part of the research focus "ICT for Energy Efficiency" of the Department of Computing Science at the University of Oldenburg, we are researching and developing new methods and tools for evaluating, analysing and increasing the energy efficiency of ICT systems, from microelectronic components to system and application software to distributed computing environments, in coordination with corresponding projects at the OFFIS Institute.

At over 55 TWh per year, ICT systems already account for around 10% of Germany's electricity demand, and this figure is rising sharply. This demand could not be met by all the wind turbines in Germany. In addition to these ecological reasons, technical and economic challenges are also becoming increasingly important, which require an intensive scientific examination of the problem of the energy requirements of information technology.

For example, the cost of supplying energy to ICT systems and their infrastructure (air conditioning, UPS) in Data Centres is becoming a significant contributor to operating costs. The heat loads that can be dissipated by air conditioning or the power that can be supplied by energy suppliers are also increasingly limiting the further expansion of Data Centres. Even at processor level, the limits of the heat that can be dissipated have been reached, which is reflected in the market by new multi-core architectures and an end to clock frequency increases.

In addition, an increasing demand is expected for high-performance, energy-efficient ICT systems that have only a limited (batch-wise) energy budget in the application context (such as mobile phones, electric vehicles, notebooks, etc.) and/or are inaccessible or difficult to access (e.g. pacemakers). In order to increase the grid-independent service life of these devices, their energy requirements must be analysed and optimised.

The focus area "Green IT: Energy efficiency in ICT" considers the design of ICT systems, including their hardware and software components, with the aim of optimising the energy requirements of information technology in a verifiable manner and in coordination with competing quality requirements (reliability, performance, etc.).

These considerations for optimising energy consumption relate to various ICT systems such as Data Centres, cloud computing, distributed sensor networks, (distributed) hardware and software systems, hardware components, software components and application software. Based on the current status of energy analyses for these individual ICT systems, it is necessary to provide models, which can also be cross-system models, to map the respective energy consumption. Based on these models, procedures for analysing and measuring the respective energy consumption must be developed and applied in order to document possible optimisations in a verifiable manner. Procedures and techniques for optimising energy consumption must also be developed for these ICT systems.

This results in scientific challenges for the following aspects, which are to be investigated for the various ICT systems:

• Modelling the energy consumption of ICT systems and their components: The analysis and optimisation of the energy consumption of all ICT systems, taking into account the load and environmental parameters, requires suitable abstractions of energy consumption using appropriate models. While such models already exist for individual microelectronic hardware components, they need to be developed and tested for cooperating hardware components in Data Centres, clouds, sensor networks and for software systems. Due to the architectural interrelationships of these systems, existing hardware models should be used as a basis.
• Analysing the energy consumption of ICT systems and their components: The targeted optimisation of energy consumption requires comprehensive knowledge of working methods and the different energy consumption levels for each task. For example, analysing the energy consumption of application software requires not only consideration of the basic energy consumption of the hardware used, but also a mapping of individual user behaviour to the actual energy consumption of the combined hardware/software system. Appropriate dynamic software analysis methods, for example, need to be developed to analyse energy consumption.
• Measuring the energy consumption of ICT systems and their components: Reliable measurements of the individual energy requirements of the ICT systems under consideration are required both as a basis for analysing energy consumption and to prove the success of optimisation. In addition to the simple measurement of power consumption "at the socket", methods are needed that measure the energy requirements of individual hardware and software components as far as possible. On the one hand, these methods are based on existing local measurement methods, but must also be supplemented by distributed measurement methods, e.g. in smart grids (see SmartGrid: ICT integration of decentralised systems), especially for distributed ICT systems.
• Optimising the energy consumption of ICT systems and their components: Increasing the energy efficiency of ICT systems ultimately aims to demonstrably optimise the energy requirements of the systems under consideration here. The aim here is to provide methods and processes that significantly reduce the current energy requirements of the respective systems. At the same time, design principles and techniques must also be developed to support them, which are aimed directly at the (initial) development of energy-efficient systems.

Research activities on energy efficiency for the individual ICT systems can build on preliminary work that is at different stages of development. Depending on the status of this preliminary work, the modelling, analysis, measurement and optimisation of ICT systems should be examined and researched in greater depth at different levels of granularity. Through close cooperation in the "Green IT: Energy Efficiency in ICT" cluster, the synergies from analysing the energy efficiency of different ICT systems at different stages of development are to be exploited and, in particular, the efficiency of integrated ICT systems improved.