Software Defined Networking And IoT Analytics

With all the talk about about edge computing, intent-based networking and IT marvels, it’s easy to forget that a formerly-great advancement, Software Defined Networking, is more than holding its own in the new age of hyper-technology. In fact, SDN is growing nicely into a major, enabling technology that fits well with today’s, and tomorrow’s, needs for flexibility in handling growing massive data volumes in fast-changing network architectures.

An SDN architecture works its magic through virtualization, creating a control layer that’s independent from the data plane of the underlying network infrastructure. This lets network engineers adjust and control network behavior without having to make changes to the physical network components.

The value of SDN is well known by corporate and data-center network managers, where SDN has been delivering efficiency and agility benefits to enterprises for several years. And SDN has grown in scope, with a spin-off technology, SD-WAN, now bringing the same benefits of virtualization to large, multi-protocol telecommunications networks.

Adding IoT Traffic

Software defined networking is a natural fit for handling the fast-growing and far-reaching data demands of future IoT devices. SDN (and SD-WAN) networks will make it easier to reconfigure network resources for sudden increases in IoT traffic, for instance, or to quickly work around an outage.

As IoT volumes increase, SDN’s ease of scalability will help data centers – and cloud vendors – provision growing fleets of devices with relatively minimal hardware investments. In fact, some cloud vendors let customers select SDN as an added service; experts expect most all cloud vendors will make it standard in the near future.

Whether they realize it or not, customers of communication service providers also stand to benefit from the efficiencies and self-healing properties of virtualized telecommunications networks. Lower network costs translate to lower consumer prices, and virtualization helps in maintaining service by bypassing failing components.

However, the very nature of SDN, its ability to virtualize components, also tends to make network underpinnings less visible to managers and management software. And this may be exacerbated by the added pressures of IoT, as event streams ramp up in volume. SDN is able to automate network provisioning and load balancing, but it’s not able to foresee the impact of such moves, or of network outages, on customers or customer groups.

To do that, and to be at its best, SDN needs an IoT analytics platform that can create a plane of visibility over the entire network.

An SDN Visibility Plane

The IoT is generating a growing flood of low-level network data from network probes, IoT sensors, customer phones and other devices. An IoT analytics framework can process that data to rule out false positives and cut out any other noise, then correlate results with contextual knowledge of everything from weather changes to the habits of customers.

It uses machine-language and other models to find relevant patterns in the data, then displays these patterns via user-friendly drill-down dashboards to analysts, network engineers, customer-service managers or others. The result is a visibility plane that sits atop the network, and that can help network managers understand the impact of network events and SDN policies.

The more advanced the analytics, the further they can go in helping determine what the best workarounds will be, or even in predicting where an outage is likely to occur. They do this by consulting historical data to show things like past performance or diagnoses or service histories of specific components. In the same way they can also predict how multiple components might act when they are grouped together.

Over time, the analytics platform builds on this information to create more informed models and better analytic insights as it gains more information.

Adding The Human Touch

It’s near impossible to tell how IoT devices and applications will act in the future, what information they’ll supply and what controls they’ll need to maximize performance. For that reason, the analytics framework should include a development environment that facilitates the involvement of teams of business analysts, network engineers and data scientists in creating new analytic applications.

Such a development environment should support the industry-standard and open- source programming languages that most data scientists are familiar with. It should support component-based application design so application modules can be reused in multiple applications. And it should employ graphics that facilitate communication among technical and non-technical team members.

Finally, it should be robust enough to handle the ever-more-complex analytic scenarios that will be demanded as organizations undergo the transition to all-digital operations. Just as IoT analytics create a visibility plane over an SDN infrastructure, they are also being used by increasing numbers of organizations to bring order to business operations; to manufacturing machines, supply chains, production planning and warehousing, and to the business process that run them.

“Digital operations” is the new corporate nirvana, just as virtualization and SDN were the new networking wonders of a few years ago. It’s important that organizations be prepared for the real-world demands of digital transformation, and that preparedness will come in large part from an effective IoT analytics superstructure.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>