IT-Zauber
"Operational optimisation of energy-efficient data centres through the use of digital twins". The goal of the IT-Zauber project is to develop digital twins of data centres, which are used to optimise the planning and operation of data centres. A major contribution is that the computing infrastructure, energy supply and cooling are considered as a whole system.
Contact
Digitalisation is finding its way into more and more parts of society and science. But the benefits can only be realised if powerful data centres are available to process the IT workloads involved. This makes data centres a must-have for innovation and progress. At the same time, planning and operation should be both cost-effective and climate-friendly. To achieve this, the focus has so far been on the energy efficiency of individual components and how they are interconnected to form an overall system. The evaluation of a data centres usually only takes into account the energy consumption of the entire system.
Unexplored approaches to optimise the operation and planning include better use of waste heat, more accurate selection between different types of processors that have different cooling and energy requirements, as well as operational flexibility, i.e. shifting computing power to times when energy prices are low or waste heat can be but to better use.
The IT-Zauber project recognises that these approaches can only be realised if the data centre is considered as a whole. To implement this, a digital twin of the data centre is created. It contains all information and data for representing the state of the data centre, as well as upcoming computing workloads and boundary conditions, such as outside temperature, energy prices or heat demand.
The digital twin can be used in two ways. For planning, one can use the twin evaluate and compare different options for expanding an existing data centre or planning a new data centre. In operation optimisation, the digital twin is an image of the current state of the data centre at any time. It can then calculate the optimal operation, i.e. how to regulate the cooling systems or how and when to distribute upcoming workloads to the computing infrastructure.
In the project, ACS is mainly concerned with creating the data model. The data model determines how the data stored in the digital twin is to be interpreted. From simple conventions of which units of measurement are used, to complex contexts such as which computing infrastructure is cooled by which cooling circuit. All programmes that interact with the digital twin can rely on the data in the digital twin to follow the data model, but must also ensure that when they pass data to the digital twin, it follows the data model.
Acknowledgement
Grateful acknowledgement is made to BMBF (German Federal Ministry of Education and Research) for providing financial support under the grant agreement no. 16ME0614K
Acknowledgement
Grateful acknowledgement is made to BMBF (German Federal Ministry of Education and Research) for providing financial support under the grant agreement no. 16ME0614K