The SDN-IoT Connection

Scale and flexibility are just two factors pushing SDN and the Internet of Things together.

By Julie Knudson | Posted Apr 6, 2015
Page of   |  Back to Page 1
Print ArticleEmail Article
  • Share on Facebook
  • Share on Twitter
  • Share on LinkedIn

That historic collision of chocolate and peanut butter—which resulted in some of the world’s most scrumptious treats—is about to be repeated. But this time, it’s software defined networking (SDN) and the Internet of Things (IoT) that appear to be on a path to technology mash-up nirvana.

Welcome the SDN/IoT convergence

There are several reasons SDN and IoT are coming together. At the foundation, IoT is all about devices. Lots of devices. And Ajay Malik, senior vice president, worldwide engineering and products at Meru Networks, said they are “all trying to connect to each other and trying to share information.” The communications stemming from all those devices—nearly 5 billion of them now and an estimated 25 billion in another 10 years, according to forecasts from Gartner—will need to be managed.

On the other side of the coin, SDN is all about managing tools. “Whenever a device is trying to talk, you want to ensure that message is received by the right parties and with the right priority, and that all that flow is completely under control,” Malik said. Acting as a sort of master brain of the network, SDN helps to manage the details of those flows.

Adding to the coming convergence is another growing technology: cloud. Both SDN and IoT are powered in large measure by cloud platforms. Much of the data that will be generated by the billions of new Internet of Things devices forecast to come online in the near future “needs to be stored somewhere,” said Wendy Cartee, vice president of marketing at PLUMgrid. “The devices need cloud infrastructure, and to build a cloud infrastructure you need virtualization, you need technology like software defined networking.”

Given the sheer volumes of data that will be part of the conversation, the cloud will likely be tapped for storage, for compute, for processing. Cloud-powered data analytics may prove the most effective method for leveraging big data analytics to “be able to analyze the data and use the data in a way that helps essentially provide the function of what that device is providing,” Cartee said.

Enterprises could find their IoT device numbers booming even without the addition of new corporate-sanctioned devices. Malik pointed to Apple Watch and similar wearables, which are likely to be popping up on the network any day. “You see what happens in schools in August and September,” he said. “Every university will probably see another 20- or 25-percent more devices, and those are just some kind of watch.” And nearly all will have a cloud backend. It’s moving beyond BYOD, and then some. “Network administrators have to think open standards,” Malik said. “That’s how we’ll manage all the traffic, that’s how we will secure the traffic, that’s how we will do anything and everything.”

In enterprises where the load of devices connecting through IoT is beginning to grow (read: explode), Jason Rolleston, senior director of product management for routing, enterprise infrastructure solutions group (EISG) at Cisco, said the pressure on the network is increasing to deal with the influx of communications. Between the sheer volume of devices and the amount of data they’re likely to dump onto the network, the traditional architecture—where data is transported back to the data center for further action—may not work as well in the IoT world.

“These sensors are pretty dumb, pretty basic,” Rolleston said. “They don’t have a tremendous amount of storage, so a lot of that information has to be stored somewhere and you have to do the analysis, and that’s just a lot of flow onto the LAN.” As additional demands are placed on bandwidth, SDN helps to create a more flexible, scalable and resilient network that can handle the influx of IoT devices and their resulting data loads.

Lay the right groundwork to make the SDN-IoT mind meld a success

With many IoT and SDN deployments still in the early stages, Rolleston said there are several issues coming along that administrators should be thinking about now. One is addressing. “If you look at an IPv4 network, you have a limited number of addresses,” he explained. The standard wasn’t built to handle the huge number of addresses many enterprises are likely to have when all of these sensors hit the IT network. “We see IoT and the number of devices being that thing that [prompts administrators to say], ‘I’ve got to have an IPv6 strategy,’” Rolleston said.

It’s also vital that administrators have clarity on what types of data these devices will be putting out onto the network. “Some will be generating unstructured data such as videos—maybe it’s a surveillance camera, for example, or another camera of some kind that’s taking pictures and sending them back into the cloud,” Cartee said. Those are likely to be larger file structures. “I think it’s important to analyze how much unstructured data and how much structured data is expected to come into the cloud, and also have a good understanding in terms of what do you do with that data,” Cartee explained.

Security is another factor administrators may want to tackle sooner rather than later. As enterprises look for ways to put push compute out to the edge to help minimize the load on the network, additional security issues could arise. These devices won’t have credentials—they aren’t people on a laptop logging into the network, or a server that’s plugged into a trusted port on a switch. “How you manage that compute infrastructure and these applications and the security policies around these sensors and devices becomes a really interesting area,” Rolleston said. “It’s one that software defined networking is very well fit to handle, and this policy intent that my sensors should never talk to anything besides my edge compute platform.”

Challenges ahead

Software-defined networking and IoT are both still young. Because neither is mass deployed today, administrators could be lulled into assuming they can think about these issues (and where they intersect) later. Malik said that’s a mistake. “I think today they’re okay and they’ll probably be okay for part of tomorrow, but then they’ll have problems because they will not be able to support the scale of needs or type of IoT devices.” Today’s environment is essentially a transition stage, and rather than stick with the old methods of using scripts and proprietary mechanisms, administrators need to be thinking about what the future holds.

Cartee said customers are already talking about accessing IoT data from anywhere in the world, and from nearly any device. “Most of the IoT devices have a mobile app associated with it,” she explained. And more to the point, a mobile app that relies heavily on the cloud. People can now close the curtains in their apartment from half a world away, all through an interface on their smartphone. That kind of control has become an expectation for many users, which Cartee predicts could translate into a challenge for enterprises. “From an administrator perspective, they need to think, ‘How do I make that work? How do I make it so seamless and so easy so that people can adopt technology and be able to have that ease and simplicity of controlling the devices with a few touches on their mobile device?’”

Comment and Contribute
(Maximum characters: 1200). You have
characters left.
Get the Latest Scoop with Enterprise Networking Planet Newsletter