By Danny Bradbury 31 August, 2018
Attention network admins: If you found the ‘bring your own device’ (BYOD) culture challenging, you’d better sit down. The Internet of Things (IoT) is about to make your life a lot more difficult.
Admins had enough trouble managing equipment that was entirely under their control. Then, employees began smearing personal mobile phones, tablets, and laptops all over their immaculately groomed enterprise networks. They were forced to balance user convenience on one side and security and manageability on the other. They dealt with that problem using a mixture of policies and technical controls.
The IoT is about to make that challenge look relatively simple, presenting a whole new set of network device management problems.
Organizations are going to implement IoT at a massive scale, an order of magnitude larger than they have done before. When you have a million devices on your network, the problems you face grow proportionately.
Spotting rogue devices
Let’s take network security as an example. How can you ensure that, among these millions of devices, compromised units aren’t subverting your network? What about rogue devices, inserted to intercept data or send fake signals?
In an ideal world, every IoT device would have a unique private key, used to identify itself to the network and prove its legitimacy. Some IoT management solutions enable deployment teams to use these keys, but the more devices there are, the more complicated and time-consuming this becomes.
This scale also creates challenges around capacity planning. How can deployment teams ensure their infrastructure is ready to handle all the data these devices are producing? The field network must be able to handle the bandwidth, and potentially the latency, in time-critical applications.
At the back end, the server infrastructure must shoulder the burden of all that data, which could run into gigabytes or even terabytes. The processing and I/O overhead could be huge, and if server-side applications are streaming that data for real-time analytics, it could be even more challenging.
Devices may not always send their data back to headquarters. In some cases, it may not be possible to process data from an IoT sensor by submitting it for back-end processing because the latency requirements of the application won’t permit it.
A sensor may monitor vibrations in an industrial turbine to see if it is about to fail. It may have to decide within milliseconds whether to shut it down or risk a catastrophic failure ripping the equipment apart. In that case, a round trip to the cloud may not produce a decision in time. It may only be feasible to run that vibration data through a machine learning model locally, deciding on the spot whether to shut it down or not. Understanding where that processing will happen is an essential part of the IoT planning story.
Adapting network management practices
Companies must also contend with problems at the other end of the scale: While deployed in unprecedented numbers, the devices themselves are tiny. They are often low-powered, without much onboard processing capacity, yet they are still active components—and network management practices must support them. Administrators must configure them to report their status to management tools, and potentially control them too.
Coordinating these devices requires new skills. While operators may be used to handling the CFO’s iPad, probing an air quality sensor on a bus that may intermittently lose its connection will probably take them into uncharted territory.
Let’s not forget the communications infrastructure and formats these devices support. The CFO’s iPad connects to your Wi-Fi network and sends HTTP traffic. The 20,000 sensors you just bought for your smart street-light infrastructure may use something else, like Wi-SUN, LoRaWAN or NB-IoT, at the physical layer. Further up the stack, they may rely on protocols like MQTT, XMPP or ActiveMQ for telemetry. Maybe they’ll connect to a broker in a publish/subscribe mechanism, which will, in turn, require a rethink about how your applications collect that information.
As you can see, the network device management considerations for IoT extend from the bottom of the stack to the top. At each layer, network designers will face an array of IoT management choices that may have knock-on effects for the rest of their infrastructure. If you thought BYOD was like herding cats, then consider IoT more like beekeeping. Be sure you don’t get stung.
- Edge Computing: The MSP’s ticket to the Internet of Things
- Security Basics: Why managing IoT security is a top priority
- The Internet of Old Things may kill us all
- BYOD: Bring Your Own Destruction!
- Five BYOD best practices for MSPs and their customers
Danny Bradbury has been a technology journalist since 1989. He writes for titles including the Guardian newspaper, and Canada’s National Post. Danny specialises in areas including cybersecurity, and also cryptocurrency. He authors the About Bitcoin website, and also writes a regular blog on technology for children called Kids Tech News. You can follow Danny on Twitter at @DannyBradbury
Original Article : Solarwinds Blog