Balancing the resources of the High Level Trigger farm of the ATLAS experiment

The ATLAS High Level Trigger (HLT) is organized in two trigger levels running different selection algorithms on heterogeneous farms composed of off-the-shelf processing units. The processing units have varying computing power and network connectivities. The ATLAS working conditions are changing due to the continuous optimization of the LHC operations and the consequent trigger adjustment. In addition, the rolling expansion and replacement of the HLT hardware changes the HLT farm composition. Therefore, balancing the available resources is essential for optimizing the HLT farm exploitation. In this paper, a tool for managing the HLT resources will be presented. The tool allows for showing, modifying and generating the HLT farm configuration, keeping the resource balance, in terms of computing power and bandwidth, across the farms under control.