Karlsruhe Institute of Technology
Knowledge of the internal behavior of applications often gets lost over the years. This paper can arise, for example, from missing documentation. Application-level monitoring, e.g., provided by the researcher, can help with the comprehension of such internal behavior. However, it can have large impact on the performance of the monitored system. High-throughput processing of traces is required by projects where millions of events per second must be processed live. In the cloud, such processing requires scaling by the number of instances.