Advanced visualization system for monitoring the ATLAS TDAQ network in real-time

摘要

The trigger and data acquisition (TDAQ) system of the ATLAS experiment at CERN comprises approximately 2500 servers interconnected by three separate Ethernet networks, totaling 250 switches. Due to its real-time nature, there are additional requirements in comparison to conventional networks in terms of speed and performance. A comprehensive monitoring framework has been developed for expert use. However, non experts may experience difficulties in using it and interpreting data. Moreover, specific performance issues, such as single component saturation or unbalanced workload, need to be spotted with ease, in real-time, and understood in the context of the full system view. We addressed these issues by developing an innovative visualization system where the users benefit from the advantages of 3D graphics to visualize the large monitoring parameter space associated with our system. This has been done by developing a hierarchical model of the complete system onto which we overlaid geographical, logical and real-time monitoring information. This article briefly describes the network monitoring framework and introduces the system's visualization challenges and previous limitations. The functionality, design and implementation of the 3D visualization system are then described in detail, with a focus on the model design, user interaction, navigation mechanisms and methods used to achieve scalability when rendering and updating many objects in real-time.