Issue link: https://www.e-digitaleditions.com/i/1476549
18 I ICT TODAY Cloud-native 5G networks will create a tidal wave of monitoring data. 0 1 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 FIGURE 1: 5G monitoring data tidal wave. THE BIG DATA TSUNAMI Before digging into the new challenges that cloud-native networks introduce, it is beneficial to discuss the issue of data in 5G SA networks—specifically, the volume of data that is expected to be generated by 5G service assurance solutions. Up to now, service assurance has been the exception rather than the rule. Most services were typically offered as best effort and service impacting issues were detected by the dedicated, physical hardware systems or worse yet, by customers calling to complain. Monitoring service quality was something reserved for services with SLAs associated with them. This has changed with cloud-native 5G SA. By abstracting the cloud infrastructure away from the 5G network and service layers, the ability to associate hardware issues with service quality issues has been broken. And even if an association is made, the net- work and service orchestrators, responsible for making sure their respective layers are performing optimally, will have made changes to these layers, rendering the previous understanding of the topology invalid. On the plus side, because the network is now virtual, the cost of establishing service visibility with active or passive probing solutions has been all but eliminated as these can now be containerized, software only functions incorporated into the service blueprints from the begin- ning. In theory, it is now possible to monitor any service at any time. For the first time, CSPs now have the tools to detect and react to any service degradation—possibly even before the customer notices. However, this level of monitoring does come at a cost. As detailed by a tier 1 North American MNO, 5G SA service assurance could generate up to 40 PB of monitoring data per hour! Transporting, storing as big data, and processing this amount of information, even in a cloud-native network, would not be cost-effective. Operators would effec- tively need to build a separate network just to carry monitoring data. The cost alone to transport and store that much data in a cloud network would break any business case. See Figure 1. What the cloud-native network needs is an intelli- gent service assurance solution, one that minimizes the amount of data being collected, scales or adapts itself up or down to address QoE issues in real time, and leverages machine learning (ML) and artificial intelli- gence (AI) to automate the entire process and reduce MTTR. According to a Q1, 2022 Heavy Reading survey, isolating the fault domain and having the right tools to diagnose these faults in cloud-native networks is the biggest challenge currently facing operations teams. More than half of the MTTR period is taken up by simply finding and identifying the root cause. By distributing intelligence within the network, ML can detect anomalies at or close to the source. In doing so, only the data related to the anomaly needs to be considered. The remaining data simply indicates that "everything is working as expected" and can be effec- tively ignored. This alone is likely to eliminate most monitoring data that needs to be moved and analyzed. When issues are identified, a centralized assurance orchestrator can quickly scale up additional monitoring and testing to rapidly identify the root cause. Once issues have been corrected and the fix verified, scale back the monitoring to its original level. The key lesson is that cloud-native networks need cloud-native and cloud-aware monitoring solutions, supported by ML and AI to deliver visibility and actionable insights at the lowest possible cost.