AI-POWERED CMDB HEALTH AND AUTONOMOUS SERVICE GRAPH RECONCILIATION: GRAPH MACHINE LEARNING FOR CONFIGURATION DATA QUALITY IN LARGE-SCALE SERVICENOW ITOM DEPLOYMENTS

Abstract

Although the Configuration Management Database (CMDB) is the key data foundation for modern IT Operations Management (ITOM) platforms, a large enterprise-size CMDB can be plagued by significant issues such as: inconsistent attribute mapping, stale dependencies, duplicate configuration items (CIs), and orphaned records. These integrity challenges have a detrimental impact on incident management, change impact analysis, service mapping, AIOps correlation and security operations. Large scale ServiceNow ITOM Deployments require a CMDB health and automatic reconciliation solution powered by AI. The proposed system consists of ServiceNow ITOM Discovery, Service Graph Connectors, Common Service Data Model (CSDM), techniques of graph machine learning for automatic CMDB quality management. It integrates entity resolution algorithms using transformers for duplicate detection, Graph Neural Networks (GNNs) for relationship anomaly detection and reinforcement learning–based reconciliation for autonomous CI remediation and governance. Previous testing in enterprise scale ServiceNow environments has demonstrated measurable improvements in the accuracy of relationships between CIs, CI completeness, reduction in CI duplication, reliability of CI to incident relationships; and ongoing, autonomous monitoring and remediation to maintain a healthy CMDB.