Sep 06, 2016The complexity of the Internet of Things means that only something as centrally manageable, scalable and flexible as software-defined networking (SDN) has a shot at dealing with the overwhelming data flows being generated by IoT devices. In fact, the prolific increase in the amount of traffic and devices is a key motivation for the rapid development and deployment of SDN technology.
Sheer scale is probably the biggest challenge of the IoT: Gartner estimates that devices connected to the IoT will number more than 20 billion in 2020, by which time the IDC predicts that 10 percent of all data on the planet will be generated by IoT devices. Service providers of all types will need to be far nimbler in their capacity to respond to changes—they'll need the ability to adjust rapidly and seamlessly by scaling up (or down) network services in certain geographies versus others.
The wide variety of devices is also an area of concern. Every device that plays a role in the IoT needs to be managed and supported. This includes legacy network devices, such as routers and switches, as well as other devices, such as drones, wearables and implantables that do not contain adequate memory to support common encryption techniques. Add to this list devices that we haven't even invented yet. Plus, these devices need to be supported without interruption, especially when they're used for safety or commerce applications. The proliferation of the IoT means increased expectations. Poor performance today already affects everything from a user's daily interactions (think: Facebook, Twitter, Gmail and so forth) to the ability to communicate (voice-over-IP services like Skype, Hangouts, WhatsApp, etc.) to the global economy (electronic payments, transfers, stock market and so on). A growing number of software and devices are being added to the system at every moment.
Why SDN for the IoT?
SDN provides the only viable, cost-effective means of managing the IoT, securing the network, and maximizing the performance of applications and analytics. Due to the IoT's explosive growth and the chaotic nature of the public internet, some of this traffic will need to be moved onto private, dedicated pathways. Otherwise, traffic will create latency issues for business-critical communication and applications. Service providers and their customers can utilize these dedicated paths to transmit business-critical traffic, while still relying on the public internet for less critical traffic, but this requires the kind of protean response that only SDN technologies can provide.
SDN—Simplicity for the IoT
SDN allows for the automation and centralized management of networks that can be remotely provisioned and managed. Long gone are the days of sending people out to colocated data centers with switches under their arms and a configuration file on a USB stick. The IoT will demand that providers rapidly onboard, configure and reconfigure devices across the globe, so having a centralized management point will become a mandatory component of that solution.
SDN—Scalability for the IoT
From a centralized point, it's possible to create automated protocols that allow for the macro-management of data flows. Augmented reality, targeted advertising, threat assessment and mitigation, and even standard daily navigation, require that network devices respond automatically and dynamically. With SDN, you can pre-define policies that prepare a network for new devices before they're even attached to the network, which basically allows you to predict IoT network attachments and be fluid in response to these new devices. SDN allows for the rapid and easy addition of new IoT devices due to its inherent scalability. Programming responsive protocols to new additions to the network means that the network can expand (or contract) as needed. This dynamic response system drastically reduces the risk of the IoT.
SDN—Traffic Flow Optimization for the IoT
The virtualization of SDN components allows for the dynamic reconfiguration of network devices and traffic, as well as automated provisioning and deprovisioning of bandwidth. So even as the IoT traffic grows, it's possible to dedicate bandwidth for high-traffic moments, or to prioritize applications involving health and safety. The global nature of the IoT means an influx of data, but with that information will come analytics leading to smarter, more predictive automation.
If devices were more aware of each other, traffic issues could be automatically resolved since they would no longer depend on manually orchestrated solutions.
SDN can also increase network visibility. Consider a scenario in which, when the internet backbone's main data routes from Los Angeles to New York City reach 80 percent utilization, network administrators prioritize alternate paths until utilization falls below 50 percent. This requires the forethought to know that a particular circuit might reach capacity, at what utilization new capacity needs to be brought online in order to avoid impacting the network, and the available dark fiber on which to light up one or more new paths.
But imagine if that same scenario were automatically resolved with myriad switches across various service providers. They could route low-priority traffic from Los Angeles to Dallas to New York City, which would free up capacity on the direct Los Angeles to New York City link. Then the traffic could be shifted back once the load decreased. This scenario could be any group of cities around the globe and, due to the chaotic nature of the internet, might need to cascade where multiple classes of traffic need to be dynamically rerouted. These scenarios cannot be hypothesized and planned for ahead of time—every eventuality just isn't possible by humans. Global automation, driven by monitoring and SDN, brings this much closer to reality.
Jay Turner is senior director of development operations at Console, a global software-defined networking services provider, and the project lead for CloudRouter, an open-source project for network collaboration and innovation. Turner has more than 20 years of leadership experience in open-source software, encompassing development, quality risk assessment, operations and support.
