The rise of Internet connected devices presents opportunities to harvest the data and leverage it for new types of intelligent applications and services. This post reviews some of the factors impacting demand and supply of a new wave of analytic applications.
Predictive Maintenance Applications and Services
In a recent visit to Mexico City for an IDC Big Data and Analytics conference , I spoke to a CIO from a large mining company. The maintenance of the equipment used in mining operations is a high value process. Unless the drilling equipment is functioning, mining operations stop. Given the distributed, remote sites where the mining is conducted, an unplanned failure of operational equipment can take days to repair, and can lead to millions of dollars of lost revenue.
He already had an existing relationship with the manufacturer of the equipment (in this case, Caterpillar), for which he paid for a break/fix service that would repair a machine after it breaks down. More recently, he signed on to an enhanced service for predictive maintenance in which the drilling equipment manufacturer provides advance warning of potential equipment failure. The result was the prioritization of maintenance spend to where it is most needed.
A month later, I spoke at an IDC Big Data event in Santiago, Chile, another country where natural resources play a significant role in the economy. I asked a group of senior IT managers where their priorities were in spending on analytics projects. I found that the interest in analytics about things (machines, physical assets) was significantly greater than the plans for analytics about customers. Here the value of advance intelligence on potential asset failure was clearly recognized. Maintaining expensive production equipment to ensure their continued remote operations is a strategic business function. Companies had taken the first steps to instrument the machines so that they could be monitored at a central company office. Now making data available for predictions to drive preventive maintenance is seen as the next critical step.
A common theme across these conversations is the growing use of predictive analytics on data from connected machines ("things") to drive new levels of optimization. A logical initial target is asset optimization. Companies were divided, however, as to whether the predictive maintenance models would be created in-house or by the manufacturer who already services their equipment. I heard two different opinions:
- In-house built: The advantage of building the models in-house (often with some professional services help) is to build up a competitive advantage (for reliable production), and also to address heterogeneous machines from multiple manufacturers.
- Manufacturer provided: The advantage of looking to the manufacturer is that (from a technical perspective) they know how to monitor and analyze data from their machines, and (from a business model perspective), they see their role increasingly as a provider of services based on their products.
Given the lack of standards in collecting, integrating and accessing machine data, the hurdles for in-house development are daunting. That favors the "manufacturer provided" approach. Yet customers will demand maintenance services that cover all machines of a certain type (pumps, drilling machines), so the services will need to expand to support heterogeneous devices.
These two conversations illustrate that software is becoming the key to value for physical things: impacting how products are designed, built, deployed, and maintained. Consider the following:
- Software-defined: Software-defined IT infrastructure enables a new level of virtualized management and optimized resource utilization. IDC now tracks these new software categories, software-defined storage and software-defined networking, as functionality moves to the software layer on commodity, rather than custom hardware platforms.
- Specialized platforms: GE Predix is a platform (using Pivotal data management technology) to monitor, integrate, and analyze asset data, which is being specialized by GE divisions and their partners. An innovation is the ability to define a software-defined machine, providing a layer of abstraction to enable value-added application development.
- Connected home: NEST (acquired by Google) goes beyond the ability of mobile-based HVAC apps that monitor and control temperature. This app learns about a household's usage patterns and can recommend settings to conserve energy use and reduce costs for homeowners.
Software is becoming present everywhere, from things in the home to the data center to field operations. Figure 1 illustrates this phenomenon, noting that innovative industry solutions are being developed and deployed for what IDC terms the "third platform":
Who has the leverage given the growing opportunity for these data-intensive, physical asset and operations-related analytic applications? Look to companies who are in a unique position to collect, aggregate or integrate data from a range of connected devices. In other words, follow the data trail.
Analytic Software and the Internet of Things
Analytics has moved from the back office to the front office and is now moving to distributed operations, leveraging data from connected machines to drive new levels of optimization. The billions of Internet-connected machines are the starting point. Internet of things-related analytic applications help to optimize the use of an individual asset (as in predictive maintenance for a wind turbine), as well as the optimization of processes where the assets are deployed (as in the positioning and use wind turbines in a wind farm).
IDC surveys point to asset and operations optimization as one of the leading Big Data and analytics initiatives. (See, for example, slide 5 of the IDC research presentation on "The Future of Big Data and Analytics" which showed "product or service innovation or improvement" as the leading driver of Big Data and analytics initiatives.) So the demand is there. With the clear benefit of building intelligence into machines and optimizing the processes in which connected machines are deployed, what is holding back deployment?
What is missing is the supply of industry-specific solutions that can make machines and the processes in which they are deployed intelligent. Several factors are delaying the availability of asset-intensive, industry-specific solutions in the marketplace. Consider the complexity of harvesting the machine data, the skill needed to build the analytical models and applications, and the capability to securely deploy the solutions. Platforms specialized for the development and flexible deployment of operational technology solutions will help accelerate getting more solutions to market. Security and data integration are IT skills. An operational technology platform built with IT rigor helps to bridge the traditional divide between IT and OT. The need is clear, and we will see the alignment of the supply to meet the demand.
Posted by Henry Morris, IDC Senior VP, Worldwide Software and Services Research