Using domain knowledge for targeted alerting in PBA for IT service management

Abstract

Process behavior analysis (PBA) has been applied to service quality control for decades. In IT service delivery management, PBA has been widely and successfully applied to monitor key performance indicators (KPIs) over time to identify process anomalies. However, we notice that PBA is often limited to a relatively small number of high-level indicators instead of slicing and dicing data for each KPI to find the key dimensions that highlight specific problems. This is due to the fact that without applying domain knowledge to identify relevant alerts, monitoring all possible dimensions leads to too many irrelevant alerts. In this paper, we demonstrate that, using domain-specific rules for each dimension as well as for dimension combinations, we generate targeted alerts for specific consumers. While the presented framework is generic, we demonstrate specifically how targeted alerting can be applied to the incident management process, in particular to monitoring incident volumes over time. We showcase the value of our targeted alerting mechanism using data from 18 outsourcing clients.