The internet of things has become a top-of-mind topic for many people in technology circles. The IoT growth forecasts...
I've seen vary widely -- estimates of the number of connected devices by 2020 range from Gartner's 20.8 billion to Juniper Research's 38.5 billion, with some IT vendors predicting even more. But everyone is talking a lot of devices, and a growing part of the IoT conversation involves using edge analytics and processing tools.
The first place to start in thinking about how all of the available IoT pieces can be put together and used to support business decisions is to understand what's meant by edge. When IoT is discussed, we're referring to more and more devices and, yes, "things" that have computing power and can connect to other devices. Those things, just like people typing at keyboards or using mobile phones, are at the edge of the networks they're on.
Take your car, for example. Today's cars have dozens of computer chips in them that control the engine, assist with braking, decide whether to alert drivers about potential problems and automate a variety of other operational processes. Not all of those devices need to communicate with anything outside of the vehicle. However, auto manufacturers want to collect information about the performance of their vehicles, so some data is being sent across the IoT for that purpose.
In addition, connected-car companies offer devices that collect data from the onboard diagnostics port to send to drivers looking to monitor their habits behind the wheel. Auto insurers do the same to fuel usage-based policies that give discounts to safe drivers. And to add a new wrinkle, autonomous vehicles might be less dependent on a driver, but they'll be more connected to analytics systems in order to closely monitor their operations and variables such as traffic, road conditions and speed limits.
A question of location on IoT analytics
One of the problems faced by organizations looking to take advantage of IoT is figuring out exactly what data being generated by devices they need to access, and where the data should be processed and analyzed. At this point, it seems as if that decision is being made primarily by technologists. "I need this data in order to ensure that the device is acting properly" is a statement often made by technical people. But it isn't the only need that should be driving such decisions.
The good news is that IoT data can be simplified and streamlined for many business analytics uses. Aggregated information is often what's needed for informing business decisions, including analyzing IoT data in an effort to improve manufacturing, distribution, sales and other business functions. In such cases, all of the raw data pulled out of IoT devices need not be sent back to headquarters; instead, edge analytics systems can be used to process and filter the data, do some analysis on it and pass on aggregated data sets to a centralized analytics platform.
While all IoT data doesn't have to leave the network edge, it's clear that the exponential growth in the number of devices being connected does mean that network traffic is going to expand significantly. What's missing is a clear line of communication between networking hardware makers and vendors of business intelligence and analytics tools.
Traffic planning smooths rough edges
The former talk about how edge analytics can eliminate much of the expected traffic, but that's based only on a technical analysis of the basic performance of IoT devices. The real network impact remains in question. And it's more than just a traffic issue. If the edge nodes closest to the IoT devices need to be more powerful in order to aggregate and analyze both technical and business information, there's also a monetary issue that will affect your infrastructure development and support costs.
The upshot is that the CIO of any organization looking at IoT must make sure that the IT and business sides are talking in far more detail -- and much earlier in the process -- than they often do. It isn't enough to have the IT team come to business analysts and say, "Here's the information that's available. Which parts of it will you need?" Instead, the analysts need to be involved in deciding what information should be captured from IoT devices in the first place.
Most people understand that the IoT will increase data volumes, but my take is that it will increase them to even larger amounts than expected -- if done properly. That means business and technical needs must be defined in concert, and IT must prepare for heavier network loads both at the edge and within the data center. Words to the wise: Plan early and minimize the pain.
Visa charges up use of IoT data tools
IoT and business analytics cross paths with new tools
How Whirlpool's CIO approached the IoT