Audio-visual Cues for Cloud Service Monitoring

When monitoring their systems' states, DevOps engineers and operations teams alike, today, have to choose whether they want to dedicate their full attention to a visual dashboard showing monitoring results or whether they want to rely on threshold-or algorithm-based alarms which always come with false positive and false negative signals. In this work, we propose an alternative approach which translates a stream of cloud monitoring data into a continuous, normalized stream of score changes. Based on the score level, we propose to gradually change environment factors, e.g., music output or ambient lighting. We do this with the goal of enabling developers to subconsciously become aware of changes in monitoring data while dedicating their full attention to their primary task. We evaluate this approach through our proof-of-concept implementation AudioCues, which gradually adds dissonances to music output, and an empirical study with said prototype.