The Internet of Things is arguably the most complex undertaking the world has ever seen. We’re talking about wiring up literally trillions of devices spread across the four corners of the globe and then corralling all of that data so it can be subjected to cutting-edge analytics to produce both macro views of reality and highly personalized digital experiences.
This is bigger than the pyramids, bigger than the Panama Canal, bigger than the moonshot.
The key element in all of this, as usual when dealing with data, is networking. In order to work properly, the IoT will need a coordinated, cohesive networking environment from the data center to the cloud to the plethora of devices out on the edge – negotiating all of the systems, protocols, topologies and other constructs in between.
The only way to realistically accomplish this is through abstraction, says Meshdynamics’ Francis Dacosta. For one thing, the extreme number of edge devices is too much for even IPv6 to handle, and the fact that most of these will have little to no memory or processing makes them difficult to incorporate into current software defined networking approaches. At the same time, there is the legion of M2M networks running on specialty program logic controllers (PLCs), which means they are largely isolated from the IP ecosystem. By abstracting all of these network topologies, we can at least entertain the possibility of building an integrated IoT without having to physically rewire the entire world, but it is still a tall order requiring the cooperation of numerous entrenched interests.
The scope of the problem starts to take shape when you consider the mish-mash of solutions currently vying for just the wireless portion of IoT connectivity. As EDN.com’s Christian Legare points out, local wireless LANs can consist of Bluetooth, Zigbee, 6LoWPAN, Thread and/or industrial protocols like WirelessHART, while the wide area is gravitating toward low-power WAN (LPWAN) through LoRa, SIGFOX and other technologies. And even once you get to wired IP infrastructure, there are all sorts of ways to reconfigure gateways, hubs, concentrators and other systems away from the centralized processing architectures of today’s Internet toward the more distributed workflows of the IoT. Groups like the OpenFog Consortium (OFC) are starting to address these issues, but the scope of the problem is so large that mainstream adoption of the IoT might not happen for a decade or more.
Vendor solutions are also starting to target a unified IoT network, but it is hard to see how one company, or even several, can make it happen. Cisco recently announced a series of gateways designed to link low-power LoRa wireless devices with Ethernet infrastructure. This should allow organizations to draw data from multiple devices without resorting to high-power cellular networks, but it is still a fairly limited solution given the range of other platforms vying for LPWAN dominance.
Meanwhile, Avaya recently demonstrated its SDN Fx architecture featuring the Open Network Adapter (ONA) IoT gateway, which seeks to leverage the Open vSwitch to provide connectivity for Ethernet-equipped devices. The SDN Fx architecture is built on the company’s Fabric Connect system, which scales out to 160,000 devices. This is impressive but a drop in the bucket compared to the full IoT. So again, at some point the Avaya IoT will have to connect to a broader networking ecosystem in order to deliver the functionality that the enterprise has come to expect.
While integrating today’s topologies to suit the IoT is a daunting prospect, it is by no means impossible. The mere presence of a largely integrated global voice and data network is proof that networks can function at this scale. But the question is whether it can be done cheaply while still providing performance that is effective enough to make the whole effort worthwhile.
At the moment, there are a lot of powerful interests who see substantial dollar signs at the end of this particular rainbow, so it seems that one way or another, it will get done.