Network Service Assurance is critical for monitoring, troubleshooting, and orchestrating cloudified telecoms networks.
5G is changing the way networks are managed. Technologies like cloud and containers are changing the way Network Functions are designed. The market is moving from providing customers with commodity network performances to a personalized experience. Service Assurance needs to be aligned with these massive changes, impacting the CSPs’ business and operations.
From one end, it will have to cope with networks that will be way more complex than they used to be, with hundreds or thousands of network functions, to support network slicing. From the other end, the network topology is not going to be static anymore as these Network Functions will be dynamically orchestrated, created, and moved among multiple data centers, technologies, and locations in real-time to ensure the quality of the service provided, reduce costs and provide the personalized experiences that customers require. It will be impossible for human users to cope with these changes in real time; hence we will need to rely more and more on automation and orchestration capabilities.
Network Service Assurance will become effectively the “observing” architecture component controlling and measuring the network behavior in real-time, providing orchestration the triggers for it to know when and where to act, in addition to its traditional NOC/SOC supporting role.
This vision of Service Assurance integration to enable automation is reflected in all the industry standards, from 3GPP to ETSI MANO and TMForum. The integration guidelines, however, are still conflicted. And it is being clarified. There are guidelines from 3GPP/ETSI and TMForum, in the form of open APIs. Integration can also happen using traditional formats, like alarms and events. Due to the variety of orchestration products (ONAP, OSM, network, and IT vendor products) and the various data models, the integration today is still custom. It will become more and more standardized in the future. Orchestration is not a simple domain, with at least 6–7 layers of orchestration hierarchically collaborating in a typical CSP enterprise architecture, with potentially multiple orchestrators in the lower layers. Service Assurance will need to interact, trigger, or collect data with those sitting in the OSS layers and resource and service layers of the traditional TMForum stack.
There will be an evolution also in terms of data collection. With service assurance as the “eyes” on the network, the more data it will be able to ingest, the better. Today’s primary source of information is network traffic. With 5G and SBI encryption, traditional tapping will not be possible anymore (for the control plane), and so-called vTAP will be standard. Also, vTAP is likely the only technically viable option to cope with the dynamic network changes, hence a critical requirement for all future networks. Cloud-native probes will be required to be orchestrated and follow the network functions to optimize the data transfer between multiple data centers.
For traditional CSPs, the user plane is still monitored via traditional TAPs and will be for the near future due to the sheer data volume, which makes vTAPs economically challenging, though technically feasible. Service assurance must be ready for a switch when technology makes this a viable option where the control plane and user plane are monitored via cloud-native tapping and probing.
Additional data sources will enrich this network vision to bring more detail and nuance to the view of service assurance:
All this data will be correlated in real-time to provide timely information for the orchestration activities. It will also offer enhanced root cause analysis and troubleshooting detail to the NOC/SOC engineers to help them reduce the time to resolution for those cases that require manual intervention.
AI/ML is but will be even more vital as a critical core function of the Service Assurance platform, enabling the data correlation, pattern recognition, and prediction capabilities required for a successful Service Assurance. The complexity of the network deployment will mandate automated Service Assurance tooling to analyze and predict network behavior and take network-corrective actions.
Additional non-real-time capabilities will enable advanced network planning and historical customer analysis to improve the network design and investigate past behaviors and trends.
Service assurance must cope with legacy and future technologies to provide an accurate end-to-end view of the customer network experience and single-pane-of-glass monitoring and troubleshooting capabilities.
To summarize, Service Assurance is moving from being a monitoring/troubleshooting function in the CSPs architecture to becoming the mandatory active component required to manage future networks, increase their quality, and control operational costs.