Assurance vs. Analytics (in Virtual-Time)

Two terms are getting much attention in the context of the digital economy, and in particular telecommunications and connectivity services – namely, assurance and analytics. In some of these discussions, there is debate about the differences between these terms and which is the most critical for service delivery. Interestingly, as lifecycle service orchestration (LSO) gains prominence in integrating SDN and NFV into today’s networks, both assurance and analytics are included in the process, with neither being promoted over the other. Perhaps there’s no real debate at all, other than semantics?

To explore this a bit further, let’s start with some definitions.

Assurance

Service assurance is primarily focused on processes and procedures involving the monitoring of a network service and its underlying infrastructure in an effort to maintain a particular service quality. We can replace the term network with connectivity if one prefers, or even broaden it to include network-based services, such as Unified Communications or Software-as-a-Service. Ultimately, there is a goal of maintaining a particular service quality, followed by tools, techniques, methodologies and processes that may be very specific to a particular service that are used together to achieve this goal.

Analytics

From Wikipedia: Analytics is the discovery, interpretation, and communication of meaningful patterns in data.

Analytics is typically used in the context of data analytics, which is considered a science for examining data with the ultimate goal to make some form of conclusion. 

When comparing both terms, the former is focused on maintaining a defined service quality, while the latter is focused on drawing a conclusion based on patterns in data. Clearly one can use analytics as one of the methods to maintain service quality, so why the debate? 

There is another aspect to these two terms related to their historic implementations. Assurance has a real-time, characteristic which means that maintaining a particular service quality implies that this maintenance is on-going and significant in the present. It’s not about maintaining a particular service quality exclusively in the future but ensuring that at any point in time – most critically, now – the service quality is maintained. 

Analytics, on the other hand, has been more of a point-in-time operation, where one gathers data, stores it somewhere, and then analyzes that data to arrive at a suitable conclusion. For example, tracking utilization over an optical node in a hybrid fiber-coax plant whose growth trend indicates a node split will be required in 12 months’ time. Therefore, there’s a real-time versus historical implication. However, with advances in computing and its relative cost, some of that analysis can occur more rapidly, allowing conclusions to be drawn based on shorter time intervals, even near-real-time.

Furthermore, the nature of SDN and NFV have generated the potential for far more dynamic connectivity services and related network-based services. This results in the production of large amounts of data in a small amount of time. Data that can be analyzed quickly to indicate a required action. Such actions include the instantiation of a new function, the application of additional compute resources to a VNF, the movement of processing from the customer edge to the telco cloud, or a scale up or scale down of bandwidth. 

In this context, analytics becomes more ‘operational,’ accounting for some of the blurred lines in the assurance versus analytics discussion. Furthermore, the increasing complexity within the network brought on by virtualization and the software-ization of network functionality, as well as the increased need for inter-operator automation, places greater prominence on automating the maintenance of service quality. This puts analytics and assurance on an equal and reciprocal plane of criticality. 

And with that criticality, one must not forget the necessity of context – the ability to frame and tag data in relation to both the localized function, but also the end-to-end service and its customer so that relevant decisions, automated or otherwise, are possible. Clearly that takes scale, the insertion of service context coupled with data collection in real-time, as well as both analytical capabilities and monitoring and verification capabilities. Maybe it’s as simple as updating the term real-time to virtual-time, — that requires analytics (both traditional and modern) to support assurance and thus the monetization of network-based services that enable the digital economy.

Blogs