your123 - stock.adobe.com
Data analytics in IoT offers new opportunities for MSPs
For MSPs monitoring thousands of IoT devices, parsing the associated data stream provides an outlet for growth. MSPs, however, must first acquire the essential skills.
Partners have an opportunity to expand their IoT services into the emerging field of analytics.
Indeed, CompTIA, an industry association, has identified three areas where IT service providers can engage in the IoT market: selling IoT devices, managing and monitoring them and data analytics. Providing data analytics in IoT, however, demands specialized skills.
"This takes analytics skill and training, as well as the ability to apply data findings to business opportunities, goals and challenges a customer might have," said Carolyn April, senior director of industry analysis at CompTIA. Offering such a service "will require investment by channel firms in terms of human resources, technical and sales skills," she noted.
The channel is "not quite there yet," April added, but said data analytics is a focus area that has grown in importance across the last three years of CompTIA's State of the Channel studies.
Vivek Kaushik refers to IoT analytics as "IoT version 2.5," explaining that the channel is still figuring out how to monetize the sale of IoT sensors and devices. But Kaushik, senior vice president of client services and account management at CSS Corp., a technology consulting services provider, called IoT analytics an "an exciting space."
The company, based in San Jose, Calif., has started capturing data coming out of IoT devices "and we're able to infer and create some predictive models to help [clients meet] the business challenge," he said.
For example, CSS Corp. is working with one client to extract data from devices on farm tractors to predict a diagnostic issue and when a tractor would require service.
Similarly, the company is helping a locomotive components and operations client do predictive maintenance on railroad equipment to preempt device failure, optimize services and inventory, and update field engineers and procurement officials on when to order replacement parts.
In the railroad example, CSS Corp. has developed a six-step methodology that starts with identifying the most critical data associated with equipment for accurate predictions. Then the data is trained and tested in machine learning platform TensorFlow to build prediction models. Next, sensor data is merged with acoustics and image data to identify patterns, and then data from breaks, bearings and wheels is analyzed to identify wear and tear.
That way, CSS Corp. officials can identify uncharacteristic system behavior and resolve issues before they occur, as well as identify the best maintenance schedule and make recommendations to clients to optimize processes.
The next step is to see if the firm can bring in another layer of intelligence to see why some equipment performs better than others, Kaushik said. Beyond analyzing historical data to predict a maintenance failure, now CSS Corp. officials want to see if they can use a locomotive company's sensors and IoT data to suggest whether a different maintenance process would extend the life of a piece of equipment such as a wheelbase.
This takes analytics to the augmented level, he said. Augmented analytics includes information coming into a centralized platform from different sources, including a locomotive company's competitors, or other trains in the same company. "Maybe they're oiling [the wheelbase] two times a day and they need to do it three times a day," he said. The idea is to take best practices from what other trains are doing and moving from predictive to suggestive, which is the augmented component.
AI will play a big role in making this possible because a system becomes increasingly intelligent by consuming more data and doing more repetitive tasks, Kaushik said. "This is the journey we'll see more and more."
Data analytics in IoT: from gyms to retailers
Managed IT service provider Logically, headquartered in Portland, Maine, is not only monitoring IoT devices for a gym company but also analyzing how much bandwidth its clients use, said Joshua Skeens, CTO at Cerdant, an MSP owned by Logically.
"We can actually tell you how many times these devices have connected in the last day or month so you can see if there are repeat customers," as well as when they are getting close to reaching their bandwidth limit. In many cases, that helps clients determine whether they need a new internet connection, he said.
Besides gyms with chain locations, Logically also does IoT analytics work for sub shops, gas stations and mall stores, Skeens said.
"It's kind of scary the type of data we can get off this stuff," he said. "We're doing this to help companies be proactive."
Some customers are clamoring for this type of information, while some are not, and Skeens estimated that IoT analytics is probably only about 5% to 10% of Logically's business right now.
"Some [customers] don't understand what their business case is," he said. "They know the data is there but don't know what to do with it."
For larger customers, Logically will tell them what type of data is available, "and it's eye-opening," according to Skeens. "They start to use it to make business decisions about whether to increase wireless capacity or that they need to add more treadmills because there's not enough" during a specific period of time, he noted.
Retailers can use sensor data to determine if store displays are working or if they need to change them, Skeens said. "You can track how long [customers'] devices stood in a particular spot so we can tell if someone stood in front of shoes," he said. "It helps them redesign the layout of stores."
Channel firms face some challenges when implementing IoT analytics, from having the right in-house skills to handling complex customer needs.
The tricky part for Logically is figuring out how to charge for IoT analytics services, Skeens said. "Is it an add-on service to what we're providing, or should it be standalone?" he said, adding that "this is something MSPs deal with all the time."
Vivek KaushikSenior vice president of client services and account management, CSS Corp.
For CSS Corp., one challenge is the fact that the industry doesn't use common IoT standards yet, Kaushik said. This results in a skills mismatch because IoT sensors use different programming languages and platforms. "We struggle with training, retraining and hiring people with specific analytics skills required for this work," he said.
Another issue is that "the expectations of customers are quite high and sometimes outcomes are not there," Kaushik said. "They think [IoT analytics] will solve their problems," but he estimates that it may take two years for partners to reach their expectations.
"It's an evolving industry -- there's a long way to go," Kaushik said. CSS Corp. is also finding it challenging to build more use cases. Right now, the firm is doing IoT analytics for about 30% of its data and analytics customers.
"We work in a very limited set of use cases and MSPs like us do not have large amounts of money to put into these innovation programs," he said.
There's a lot of buzz, and every CIO and CXO wants to do more with AI, IoT and smart intelligence, Kaushik said, "but the actual investments that go into developing it are not as proportionate."