Intelligent Distribution of Computing Tasks and Workload Management
A key focus of the project is the intelligent management of workloads, or computing tasks. To this end, we are integrating software solutions that can automatically distribute computing tasks across suitable systems. With the help of an orchestrator, workloads can be distributed across different systems based on time or location. The orchestrator is responsible for the strategic planning of computing tasks and decides when and where workloads are executed. In doing so, it takes factors such as the availability of renewable energy, infrastructure utilization, and application requirements into account.
This allows applications to be executed preferentially where sufficient renewable energy is available or particularly favorable operating conditions prevail. The IT infrastructure can thus respond to signals from the energy system and use resources more efficiently than before. In addition to workload distribution, we are investigating how the power consumption of individual computing systems can be controlled flexibly.
For this purpose, we use methods for dynamically adjusting the clock frequency and voltage of processors and graphics processing units. Experts refer to this approach as »Dynamic Voltage and Frequency Scaling« (DVFS). When less computing power is required, energy consumption decreases. When demand increases, the resources can once again be made available at full capacity. In addition, reduced power consumption can improve energy efficiency and reduce the load on technical components.