Intelligence in today’s networks is both an expectation and a requirement, and that intelligence comes through analytics.

Networking has significantly evolved over time from an architecture and operations point of view. Few of the notable trends include:

• Today’s networks are no longer characterized just by physical devices that provide connectivity. What used to be a simple collection of hardware appliances and rigid pipes is evolving into software-enabled virtual resources that can be provisioned on demand and programmed centrally.
• Virtualization and cloud are changing how IT infrastructure, including network resources, are getting utilized in the data center and telecom environments.
• The distributed nature of applications poses additional dependency on the underlying network infrastructure. These applications demand that network resources be provisioned quickly in hyper-distributed environments and operate efficiently.

The common theme across these trends is agility and centralized control. These are essentially the key attributes of a software-defined environment. software-defined networking (SDN) and network functions virtualization (NFV) are technologies that can offer significant flexibility and automation within the IT and telecom landscape.

However, if not looked at holistically, these benefits can come with added complexity. For instance, the dynamic nature of creating and moving software-based functionality can pose significant management challenges. End-to-end management must cover both physical and virtual infrastructure in multivendor environments. It is also important to understand how compute, storage, and application components integrate with the physical and virtual network components.

Analytics play a crucial role in overcoming complexity and other challenges associated with this transformation. The fundamental premise of analytics is to transform data into actionable insights.

Operational analytics involves collecting, analyzing, and reporting data to improve operations. With the right set of operational analytics, continuous visibility into the network and data center environment becomes clear. Without analytics, network/IT operators can experience productivity bottlenecks and blind-spots. Visibility can be manifested by descriptive and predictive analytics capabilities in the form of hindsight, insight, and foresight. As operators rollout SDN and NFV capabilities, variety of data-sets pertaining to network faults, configuration, traffic parameters, security, etc. can be leveraged to deliver real-time insights for effective troubleshooting, planning, migration, and other day-to-day operations. This ensures optimal provisioning of the virtual network environment as well as seamless interoperability between physical and virtual resources. This becomes more compelling in a multivendor and multitechnology environment. For example, in a heterogeneous data center/private cloud, analytics can offer a layer of abstraction to overcome the challenge of interoperability between otherwise “siloed” systems.

Cognitive analytics offers a new generation of solutions that use advanced algorithms to learn directly and automatically from data. With such self-learning systems, cognitive analytics can enable valuable insights for better network prediction and optimization. In several scenarios, enhanced intelligence leading to smarter decisions is an expectation. Advanced data mining and machine learning capabilities of cognitive analytics assist in automatically detecting anomalies and patterns. This can help predict network faults and failures for proactive outage avoidance. Network operators can conduct “what-if” analysis to simulate before and after scenarios. They can ensure that the quality of experience is sustained through the SDN journey.

Other examples of cognitive analytics use cases would be network capacity planning based on anticipated growth, resource optimization, and prevention of security vulnerabilities such as distributed denial of service (DDoS) attacks.

Automation is an important consideration with a robust analytics platform. Automation boosts efficiency of network operations, especially when it comes to configuration and monitoring. This is more so the case with multivendor, heterogeneous software-based network environments. Successful implementation of automation requires deep intelligence about the health of the network, dynamic topology, and the application environment. A fully-automated, closed-loop solution would be one where the analytics platform not only provides actionable intelligence but also integrates with an orchestration platform to enforce policies based on that intelligence via a policy engine.

Clearly, operational and cognitive analytics can help network operators realize the full potential of SDN. In fact, many operators are regarding analytics as essential and integral elements of transformation to SDN.