The 7 best practices for IT-OT convergence
October 15, 2020
October 15, 2020
The well-executed integration of a company's business and operational data is critical to digital transformation success. It's also a full-fledged change effort that spans so much more than a list of "tech-tasks": successful "IT-OT integrations" needs prudent organizational change, insightful data governance, and new ways of collaborating across the business. So, how can executives ensure execution excellence?
By Simon Coombs, Accenture
It's been clear to me throughout my career … and it's clear in the research, too: companies which prioritize the well-planned integration of their IT-systems with their factory- or warehouse equipment will see better results from their overall digitalization efforts. Their peers which don't do this, on the other hand, will struggle ever to make a return on their digital investments.
So, how can any company move into the group of IT-OT integration "Champions" and steer clear of the risks which come with the—widespread—"just build and the value will come" fallacy?
Here's my answer to this very question, which I've structured by listing the seven management best practices I've found most helpful for more than 50 projects, often highly complex integration projects across a multitude of industries.
This is essential. IT-OT integrations are about driving action… and every change effort should start with the definition of the expected results.
So, be very clear about how your company's digitalization and the integration of IT and OT data will create value for your business. Then, use this clarity to specify the actions you expect the effort to drive.
I still see too many companies who missed this critical step in their projects. If data and insights are to create any value, they must influence people and systems' behavior. So, make sure you start with that in mind—that is, in all stakeholders' minds!
Only once you have clarity and alignment around the change, you'd like to see should you start to plan the IT-OT integration itself, following the steps outlined (read from right to left!):
When it comes to planning the integration, it does pay to stick to industry best practice. And the current one for viewing an IT-OT project is the ISA-95 standard. This defines IT-OT landscapes as stacks with four layers, a view closely aligned to how vendors typically explain their solutions. I've always found this a beneficial way to think about convergence projects.
The next figure shows a common starting point for many of these projects—and a common opportunity: Companies often pull together IT and OT data in their wide reports, so they're seldom optimized for deriving insights helpful to daily operations. Seeing and then tackling this "optimization gap" is how you'll drive convergence value.
The above view shows something else, too: the fragmentation of a company's OT landscape.
While IT apps are almost always standardized across the enterprise (e.g., your company might use SAP software as its EPR system, and match other software to the platform), OT solutions rarely are. Most businesses use a multitude of vendors and versions—sometimes even within a single site.
The key here is for you to embrace this chaos and not fight it.
I know, I know: the idea of sorting it all through some homogenization effort will be very tempting. But in nine out of ten cases, these efforts are too cost-intensive and risky to be worth it. Sure, reducing the number of vendors will benefit supply chains and make skill-building easier. But the necessary reduction projects often take years to complete and so the landscape will remain fragmented over that period.
Hence my preferred best-practice: Don't battle the chaos—manage it. We often use "layer 3" of the ISA95 model as an "optimization layer," i.e., to combine different IT and OT layers across the entire business. This strategy makes the OT complexity manageable while helping to reconcile the various security needs of IT and OT… but—more on that below.
For IT, the common security requirements are—in order of priority—"CIA": Confidentiality (of customer, financial and recipe information), integrity of data, and availability.
For OT, the order of priorities is different—it's "AIC": Availability—without data flows and controls the "lights go out" (like, literally); then integrity of configuration data—a critical quality control—and then confidentiality as the "lesser" of the three (even though still necessary).
Besides its fragmented nature, OT comes with another characteristic you'll have to manage—"time-series data." It's a critical piece of every IT-OT-integration effort, and a kind of data which is different from data commonly found in business IT systems. It's also time-variable, which causes a few more complications. By and large, managers will have to solve for two specific issues:
The best way to do this? Creating a "trusted data layer" where manager feed all OT "historian" data into an analytical framework. There, algorithms will enrich it with contextual metadata while also handling some essential data quality management. If configured correctly, this layer also enables data-chain governance (i.e., governance from the sensor to automation controls) and monitoring plus support functionality.
The latter is especially helpful because, even though the most IT-OT convergence solutions work well for a while after set-up, data quality issues tend to occur after a few weeks (often due to problems with the underlying operating model). If you cannot spot and fix these, the erosion of trust into the insights generated by the solution can be significant—and prevent operators from acting on them, effectively ruining the ROI of the entire convergence effort.
Speaking of acting on insights: Once your trusted data layer is in place, you can start to add "action drivers" to it, i.e.: analytics, visualizations, and alerts, which help your operation experts to find and fix incidents or issues quickly. Remember: the IT-OT integration is merely the means—not the end, meaning: You must ensure the technology feeds back into human and systems behavior.
My go-to recommendation here is to integrate your IT-OT solution back into IT-led work- and asset-management-solutions, so your IT experts can optimize workflows and interventions.
This way, you can keep human experts "in the loop"—a solution preferred by most clients I work with—and easily monitor the "business adoption" of your IT-OT solution. Both can be extraordinarily productive drivers of the changes in culture and behavior you're aiming for, and help create a pro-active, data-driven mindset within your operations staff.
Remember: you can always add some fancy solutions for fully automated intervention later.
Since I've mentioned analytics, I should also offer a word of warning: Don't plan to be doing all your analyses in a cloud-based “data lake” platform.
Sure, cloud platforms enable very cost-effective, flexible, and highly secure data analytics—especially if you're working with IT data. But once OT data comes into the mix, these benefits might not only go away but you need to replace them with disadvantages:
High volumes of real-time, time-series data—i.e., precisely the kind of data OT generates—can lower the performance of even the best cloud analytics solutions … or drive hardware subscription costs through the roof. And they can incur steep development and support costs in cases where a standardized platform must be customized to take in highly specific data.
So, make sure you and the IT experts you're working are clear on how OT data is different—and how an out-of-the-box cloud solution might not your best bet. A hybrid solution, like the one depicted in figure below, will likely be much better. Such architectures keep historian and IoT data separate and handle raw time series data and quality management tasks themselves. They only "tap" the OT cloud data in the few cases where they need advanced analytics—and usually compress their inputs before uploading them.
Once you have your architecture and infrastructure planned out, it's time to turn to the most important component for IT-OT of convergence success: organizational change.
Almost always different organizations of different people, with a different mindset (specifically: centralized IT teams and de-centralized site Operations) manage IT systems and OT systems. And if you don't plan your efforts around the particularities of this "cultural divide", you'll struggle to meet your overall goals:
The rise of IoT solutions has brought most OT solutions much closer to IT-like attributes IT-OT convergence. And as OT becomes more like IT, both organizations must change.
The clients I work with have used different strategies for managing this shift, and most of them are now pursuing a combined and centralized IT-OT organization model. Such efforts often remove direct control from plant managers' hands and put it into the hands of off-site OT experts. But given the close link between these experts' work and the need for on-site plant management and maintenance, both groups will have to maintain a close relationship. So, they should be given the clear objective and responsibility to do just that.
In most models I've seen, plant and regional operations managers have historically had their "own" OT teams, and, as a group, work closely with a local IT manager who ensures alignment with central IT.
As IT-OT convergent solutions become more important, the local manager needs to consider both IT and OT. Their local team is often combined into a virtual Center of Excellence that seeks to provide standardized and efficient solutions and data across the whole enterprise.
And as IT-OT matures, the organization becomes increasingly centralized. This shift enables the other IT components and the local roles to focus more on connecting.
These were my seven best practices for IT-OT conversion; I hope you find them helpful to your journey and project. If so, do let me know—and if you believe I've missed something or have a different point of view on what I shared, reach out and e-mail me: What's your best practice, and how could others apply it?
Accenture’s Operational Intelligence Value Framework supports this proven. We can deliver face-to-face or through virtual workshops, to find the business ideas, confirm the outcome/actions/insights, and prioritize these into use cases for deployment. For more information please go to the Digital Production & Operations page.