Real-time network & service mapping: why wait for NFV?

      2017-04-10

      Previously published on Light Reading

      The need to know what is going on in the network and how it is impacting service quality and customers in real time has grown in urgency. Active real-time network topology is a hot topic as operators move ahead with software-defined networking (SDN) and virtual network functions (VNFs).

      Structured data is currently locked inside enterprise applications -- e.g., customers, equipment, contracts, products, infrastructure and services. Network inventory data, for example, is typically held in static relational databases that are not particularly accurate or capable of handling highly dynamic network inventory data and relating it to customer site or service type.

      Operators spend billions on network technology, yet for a particular customer still cannot be 100% sure, when trying to fix a problem or anticipate the impact of a planned network change, what equipment the customer has, at what sites, what the wide-area network looks like, and how a change could impact that customer's services.

      With a move toward dynamic networks to satisfy equally dynamic and demanding customers, a reliable view of the network topology and inter-dependencies is necessary. Given the high velocity of network data, the number of devices that will be connected to the network with the Internet of Things (IoT), there is an urgent need for high quality data in order to automate provisioning of services as well as assure service performance, and proactively manage and anticipate faults.

      There are companies that can do real-time graph modeling of service providers' network topology and services. Ontology, recently acquired by EXFO, does just that. The idea is that EXFO Inc. (Nasdaq: EXFO; Toronto: EXF) probes and analytics can measure and analyze network and service performance in real time, and Ontology Systems can align data sources to determine and quickly visualize the likely cause of poor performance.

      An advantage of the graph approach is that models can be developed on top of existing operator data and multivendor inventory databases, so it's not a case of throwing out the current investment. Another advantage is the ability to map the network structure across domains from the RAN, backhaul infrastructure to local access networks. It's useful to join those disparate data sources together very quickly to get a real-time picture of complex relationships between different network or data center operations. The topology data can also be integrated with big data analytics on network elements and other performance management tools to deliver end-to-end service assurance.

      Service providers are also looking at integrating inventory topology graphs with service catalogs to have a real-time view of network and service components, across physical and virtualized networks.

      While NFV is a key driver for active network topology mapping, service providers see value in using network graphs today to do common cause analysis to speed up fault resolution and service impact analysis and ultimately to improve the customer experience. This could help to correctly notify customers of SLA breaches or upcoming maintenance rather than over-notifying them or notifying them about something that will have a negligible impact. Complex fault management is another area where graphs are being used today. Service providers can import multiple RAN failures directly from the fault management or trouble tickets to search for the predicted common causes across transmission, core or RAN networks.

      Another leading global mobile operator is integrating real-time network topology graphs with its CRM systems so service managers can understand enterprise customers' networks and services and manage the change management processes to improve customer satisfaction through accurate change impact notification.

      In addition, we also see graphs as a useful way in the future for operators to manage cloud-native networks and monitor microservices -- where VNFs based on microservices could be running in containers on cloud native architecture with standard reusable components that are highly available and hyper-scalable. Graphs and visualization will be a useful way to monitor containers and service chains of VNFs and the relationships with infrastructure and service quality, as they move around the data center and network.

      — Sandra O'Boyle, Senior Analyst, Heavy Reading