Cost, security, and how to choose a platform for the next fifteen years.
Nobody replaces a SCADA system because they want to. The application works, the operators trust it, and in many plants it is the only reliable documentation of how the process is actually run. So the decision to move it comes from outside: a support renewal priced like a new purchase, an upgrade that needs an operating system the plant cannot run, a security review that will not pass, or the one engineer who understood the scripts retiring next spring.
A decision that arrives that way arrives late, and late is expensive. Plan before the event and you choose the timing. That is the one thing early planning buys you, and it is worth this article.
I have been designing this class of software for three decades. For a long time the honest advice was to leave a running application alone, because replacing it meant redrawing every screen by hand and hoping nothing was lost on the way. That advice no longer holds. You can keep the engineering you already paid for and move the platform under it. The film shows how it is done: https://youtu.be/VSTt4nkqK1s. This article is about why, and why now.
The cost of standing still
A legacy application has a running cost, and it never shows up on one line. Put these four lines on paper for your own plant.
- What you pay to keep it alive. Support renewals, the operating system and hardware versions it requires, and the spare machines kept only because the software cannot move.
- What every change costs. On a flat tag list, adding one pump means touching tags, alarms, history and several screens by hand. Take the last three changes you made and count the hours.
- What you cannot do at all. A browser view for the supervisor, data to the business systems, an AI model you host that can read the live process. On an old platform each of these is a project of its own.
- What it costs to be late. The forcing event arrives before the plan does: an unplanned outage, a failed audit, a rushed migration at the worst possible price.
In my experience the first line alone pays for the move in many plants. The third line is where the larger return is. The first two keep a system running. The third one improves the process.
Why the move stopped being expensive
A migration starts from the application you already have. We do it in three steps.
Read. Every file is read and accounted for: tags, windows, scripts, addressing, navigation, alarms and logged points. Nothing that is running is touched.
Model. The flat tag list becomes a structure of equipment. In our example report, drawn from delivered projects and fully de-identified, 7,553 flat tags became 8 equipment templates, instanced 516 times.
Prove. The result is exercised on the running client, and every construct in the source is either carried across or logged with a reason. The same example had 8,940 constructs. 8,282 were carried and 658 were logged, each with its reason.
The screens are the part everyone expects to be hard, and they are now the easy part, because the AI builds them from the model. Rebuilt on the plant model they also come out smaller. In that example eight equipment templates carry 6,800 of the 7,553 tags, and one fix to the pump template reaches every pump. The AI does the building. What still takes judgment is the model of your plant, and that part stays with people.
More on the method: https://tatsoft.com/model-the-plant-not-the-screens/ and https://tatsoft.com/migration-assessment/
Security is part of the business case
An application that cannot leave an old operating system is a security position, whatever else it is. When you evaluate the next platform, ask it what your own security team will ask you:
- How is the code built? FrameworX is managed .NET code throughout, and its multiplatform runtime on .NET 10 deploys to Linux, Docker and edge devices.
- Who can change what? Role-based access control with granular permissions, and write operations denied by default.
- Where do credentials live, and who can see what changed? Encrypted secrets, TLS on network services, and audit logs you can export to your SIEM.
- How are vulnerabilities handled? Monthly security bulletins, monitoring of the components we bundle, and out-of-band updates for critical issues.
- Where does plant data go when AI is involved? You bring the model, running locally or on your own premises. FrameworX ships none, so where the data travels stays your decision.
FrameworX is designed to support IEC 62443 and NERC-CIP compliance programs. It is not third-party certified. I would rather say that here than have you find it in the audit.
Choose for the next fifteen years
Every generation of industrial software has a platform that gets chosen because it was the safe answer for the last ten years. Your plant will live with the choice for the next ten to fifteen. Six questions predict how that goes.
- Does runtime logic compile and run inside the runtime process?
- Is the scripting language current, and can it load today’s numerical and machine-learning libraries?
- Is a screen built once for the control room, the browser and the phone?
- Can the running system call an AI model that you host yourself?
- What does a complete system cost to assemble, and again when the plant adds a line?
- Can you get your whole application back out, in a form another tool can read?
Ask them of every vendor, us included.
The economics, last
FrameworX licenses are perpetual and the price list is public. Enterprise includes unlimited clients, and moving up an edition costs the difference between the two. Build your system on paper from the list, then put it beside what you wrote under the cost of standing still.
Where to start
Most migrations stall for one reason. Nobody wants to sign a project they cannot size. So we changed the order of the work. You send us a read-only copy of the application, from any platform, and our engineers read it back to you object by object before anyone quotes anything: the inventory, the tag model, the addressing, the alarm and history census, and a map of the navigation. The fixed price for the migration comes with that report. If you don’t accept the result, you don’t use it and you don’t pay for the migration. The report is yours either way.
On 15 October we rebuild a legacy application live, in one hour. Bring yours: a free assessment, a fixed price, and you pay only if you accept the result. Register: https://tatsoft.com/webinar-scada-hmi-migration/
Author Bio: Marc Taccolini brings three decades of industrial software experience as designer of Tatsoft FrameworX and previously InduSoft (now AVEVA Edge).
Aim High. Start Simple. Scale Without Limits.