Error Monitoring for Legacy Mission-Critical Systems

Error data collected at runtime play a key role for dependability analysis and improvement of software systems. The use of monitoring frameworks for legacy mission-critical systems is hindered by limited intervention degree and low intrusiveness requirements. We present the design and experimentation of an error monitoring service for a legacy large-scale critical system in the Air Traffic Control (ATC) domain. We describe the details of the API realized to collect both direct data (event logs, execution traces) and indirect data (system resources' utilization). We present experiments with the ATC industrial case study, showing the efficacy of combining different data sources for error detection and propagation analysis, with an acceptable overhead at high monitoring rates for such a class of systems.