Issue link: https://www.e-digitaleditions.com/i/1414667
28 I ICT TODAY Customer Impact Assessment Once a fault or performance issue has been detected, categorized, and located, ML algorithms are able to correlate it with customer usage information and contractual commitments to assess both customer and SLA impact. Elevating performance issues to the level of customer and SLA impact enables informed decisions to be made and operations teams to effectively prioritize their efforts for the desired business outcome. Free Up Operations and Technicians With the ability to automate root cause analysis, diagnose and locate faults automatically, and predict service-impacting issues, AI-enabled service assurance not only reduces MTTR and upholds stringent SLA requirements, but it also reduces the operational costs associated with manual analysis. Automating diagnosis and classification lessens the need for more senior or specialized technical analysts, thus freeing up valuable experts for more valuable tasks. Predict Not React Advanced AI capabilities allow operators to look forward, leveraging historical, internal, external, and third-party data sources to predict network resilience, performance, and behavior. When correctly applied, these techniques create models that allow identification of causality between seemingly unrelated events and outcomes. These models leverage historic and current-state data streams to probabilistically predict future issues and enable providers preemptively to mitigate service impact, thereby avoiding loss of performance or outages before they happen. For example, planned work involving fiber cables could be automatically rescheduled to avoid an over- lap with a severe weather event that has historically impacted certain data centers, anticipated spikes in traffic volumes due to an upcoming planned sporting event, and close-to-threshold packet loss along the backup dark fiber route. Advanced AI capabilities allow operators to look forward, leveraging historical, internal, external, and third-party data sources to predict network resilience, performance, and behavior. Cross-Silo Analysis The data center and DCI market is in a period of rapid growth, including frequent mergers and acquisitions. This results in providers needing to manage and main- tain not only multiple generations of equipment, age of component and diversification of equipment vendors, but also a quilted patchwork of fiber networks. This environment of operations silos and equipment frag- mentation means the creation of timely insights and effective alignment of service indicators across the breadth of the network can be a cumbersome, costly, and error-prone task. Machine learning approaches can identify patterns and similarities in performance and faults across the network. When combined with additional topological data, it can help harmonize alerts and fault discovery across both knowledge domains and network silos, presenting a one-network view to aid analysis and resolution.