Engagement infrastructure risk analytics

This paper presents the effort toward building a predictive analytics framework for engagement risk management. We propose a method to predict a number of problematic servers for unknown client infrastructure environment to calculate the risk based on historical servers’ classification. The first part of the algorithm is designed for grouping configuration items based on proposed similarity engine and predicts how many problematic servers are expected in the formerly unseen environment. Second part of the method compares results to currently maintained infrastructures and by using defined thresholds nominates them to implement proposed improvement actions. The analytics engine provides significant improvement over traditional risk assessment in capturing risky infrastructure environments when tested with real client data.