Using Tableau Server process logs and OS metrics to distinguish between a VizQL Server memory pressure problem and a Data Engine query execution bottleneck during peak concurrent load