How to tame IoT: A guide for IT Ops

As the Internet of Things (IoT) continues to grow within and across businesses, all those "things" will need to be managed. IT operations management (ITOM) can help do that, but there's still the question of how companies connect and monitor all of millions of IoT devices in a meaningful way.

The IoT's impact on infrastructure and operations (I&O) will likely be significant, and  leaders who aren't prepared will face major consequences, according to a March report from Gartner titled "I&O Leaders Must Get Involved With Current or Planned IoT Initiatives." 

Here's how ITOM is starting to help manage the IoT.

The State of Analytics in IT Operations

The time to act is now

"The number of IoT initiatives is rapidly increasing, and infrastructure and operations leaders cannot afford to be caught flat-footed," according to the report. "IoT initiatives can have consequential impacts on I&O, and I&O leaders must better first assess and elevate their IoT involvement.”

In the next few years, IT operations will likely drive, implement, or operate about 40% of all IoT initiatives, according to Sanjit Ganguli, research vice president of IT operations at Gartner and co-author of the report.

I&O teams will take a more active role in designing and managing these IoT initiatives. Consequently, organizations will rely on IT operations more frequently to have visibility into the inner workings of IoT systems as well as to analyze the IoT data that is produced.

Automation is critical

The scale of IoT will bring challenges, including the need to manage many more IP addresses. "In the past you might have needed a handful of IP addresses; now you might need hundreds. It's a huge challenge from a provisioning perspective, an automation perspective."

IoT initiatives will be severely handicapped without the use of proper automation, but many enterprises have half-baked automation strategies. Although they have automated certain areas, there’s still a lot of cultural resistance to it.

And automation vendors really haven't taken a stab at the IoT market yet because it's not particularly standardized or well-defined.

But that will change, Ganguli said.

"Where you could start to see some interesting things ... will be with the Puppet, Chef, Ansible type of toolsets out there that hopefully will start to look at IoT as yet another type of endpoint that they can start to automate."
Sanjit Ganguli

Apply deep learning and AI

Companies considering IoT strategies don't need to worry about how to monitor individual devices, because existing vendors already have monitoring management toolkits, said Clay Roach, chairman and CTO at StreamWeaver.

Rather, organizations will need to figure out a way to integrate the various IoT components with one another and with back-end systems quickly, easily, and efficiently. "We're recommending finding those tools that speed up and/or make that integration process really easy," he said.

Another consideration revolves around how to manage the volumes of data generated from IoT devices. "The number of things that need to be monitored is just increasing the overall volume of metrics being produced ... that a traditional operations team would be using to manage health and status and uptime availability," Roach said.

It will be difficult for IT operations to understand and manage increasing numbers of those devices without IT operations capabilities, such as automated baselining and anomaly detection capabilities, according to Roach.

"If you have analytics tools that actually do that learning, utilizing deep learning techniques and some of the other advanced [artificial intelligence] analytics capabilities, you don't have to teach a human operator to do that."
Clay Roach

The tooling can learn characteristics and behaviors so that you alert when something is going differently than before, he said.

Quiet the noise

The IoT explosion is creating a lot of noise for IT operations management, said Roy Illsley, principal analyst at Ovum.

IoT is quite vast in terms of what ITOM could be asked to control, including the devices and the gateways that the IoT devices are connected to, he said. So the first thing IT Ops management will help with is understanding where and what needs to be managed, because all facets won't need the same level of management or the same degree of care and attention.

The other IoT management element that ITOM can bring is visibility into the network and network traffic, he said.

"While you're managing the device in terms of making sure it's protected and it's available and it's working, you also want to be able to look at how it is communicating, whether it's reachable, whether it's delivering the service in the right time frame," Illsley said.

ITOM already has vast experience dealing with those issues. But the tools can better assist if they're just slightly tweaked to handle new devices and new communications protocols, he added. Increasingly, ITOM is using machine learning and AI as a way to be able deal with the explosion of IoT devices.

"As you scale up, the automation becomes more significant, and that points you to the fact that you need to have some intelligence built into the solutions that can make decisions for you on how to manage a standard gateway, for example."
Roy Illsley

Keep evolving

Fran Korosec, chief operating officer at BRIDG, formerly the International Consortium for Advanced Manufacturing Research, agreed that ITOM has to better connect to the plethora of devices that are available and then do something with the data.

That’s going to push ITOM systems into including more highly functional data analytics and potentially even into AI, he said. It comes down to managing more devices than what ITOM systems do today.

"How do they manage the data traffic? How do they better store data? How do they make data more accessible for AI and data analytics systems and have that data be searchable in a fast manner?" he said.

If the data can't be analyzed in a timely manner, it becomes less effective at helping IT managers make better business decisions with all the data available.

Although ITOM has been evolving to meet the needs of monitoring and managing IoT at scale, has it been evolving fast enough or in a manner that matters to IT managers? 

The answer depends on what the IT manager needs, Korosec said.

"If it’s an IT manager who cares about building automation systems to be able to adjust temperature, humidity, and pressure in a building, ITOM systems in that regard have been coming along nicely. But if it’s an ITOM system that needs to be intimately involved in a manufacturing operation at the shop floor controls, say at a robotics station, those advances are still evolving."
Fran Korosec

[ Upcoming Webinar (Oct. 23): Simplify Discovery and Change Management for Cloud and Container Environments ]