Issue link: https://www.e-digitaleditions.com/i/1476549
20 I ICT TODAY FIGURE 3: Closing the visibility gap. According to the GSMA global survey of MNOs conducted at Mobile World Live in May of 2021, with upward of 89 percent of 5G SA services expected to have SLAs attached to them, 61 percent of SLA-governed 5G revenue cannot be delivered without real-time service assurance. This problem can become very expensive, very quickly. THE EVERCHANGING NETWORK It is not just the cloud-native network and service layers that are dynamic. Surprisingly, the infrastructure hard- ware supporting these networks is also quite dynamic and sometimes overlooked when searching for the root cause of service impacting issues. Firmware revisions, operating systems revisions, container orchestration system updates and revisions, and even the swapping out or upgrading of physical servers, including config- uration parameters, can all lead to QoE issues, even when they are done properly. Cloud-native networks are extremely complex and have moved well beyond the capacity of humans to grasp, so misconfigurations and unintended consequences are to be expected. CLOSING THE GAP A couple of important questions arise when addressing closing the gap. How can this visibility gap be closed? How can network and service visibility data and the cloud and IT observability data be democratized so that both sides have the same view of the data and can work in cooperation to deliver a better 5G experience to their end customers? QoE QoS Observability metrics, traces and logs Visibility KPIs, KQIs, xDRs, traces Services Network K8s, OpenShift, … Infrastructure Visibility linked to observability Contextualized data Health indicator Correlated, full-stack, ML-powered assurance to detect, predict and prevent degradations Analytics Fortunately, there are open standards activities that are also vendor-neutral, such as the OpenTelemetry project within the Cloud Native Computing Foundation (CNCF), that are addressing the questions and issues. By extracting observability data from the cloud infra- structure and contextualizing it with timestamps, plat- form, OS, container and pod orchestration data, along with workload metrics, this OpenTelemetry data feed can be integrated into an ML/AI-based analytics plat- form and then correlated with QoE and QoS data from the network and service layers, effectively closing the visibility gap. In this way, there can be a single "pane-of-glass" view of the complete network, service, and infrastructure stack supporting 5G services, allowing for the rapid identification of root cause issues and, if applicable, the specific hardware, platform, and/or resources respon- sible for the QoE degradation. By tearing down the barriers between operations teams and democratizing the assurance and observability data, the orchestration and automation systems driving network and service performance optimization will have a complete view, leading to better decision-making. See Figure 3.