SCADA and Historians

Supervisory systems and the historians behind them. Architecture, tag standards, redundancy, and the difference between a system operations trust and one they work around.

What lives in this layer

This is ISA-95 Level 2-3: the software that lets people see and supervise more than one machine at a time, and the databases that record what the process did. On the SCADA side that means platforms such as Ignition by Inductive Automation, FactoryTalk View SE, AVEVA System Platform and InTouch (formerly Wonderware), and Siemens WinCC. On the historian side it means process historians such as AVEVA PI (formerly OSIsoft PI, also sold by Rockwell as FactoryTalk Historian), AVEVA Historian, and Ignition’s own historian modules: since version 8.3, a built-in Core Historian based on QuestDB, with the SQL Historian still available for plants that store history in a relational database. Alarm management, operator security and audit trails live here too.

Naming the layer matters, because the most expensive mistakes in modernization come from solving a problem at the wrong level of the stack. A historian will not fix bad tag data coming from the controller, and a new SCADA platform will not fix an alarm philosophy nobody wrote.

The problems we see most

Islands of HMI. Each line has its own screens, built by a different integrator in a different style, with no shared tag naming and no single place to see the plant. Supervisors walk the floor to learn what a central system should tell them.

Alarm floods. Hundreds of active alarms, most of them nuisance, so operators acknowledge everything and the alarm that matters is buried. The system is technically working and nobody trusts it.

Servers nobody owns. SCADA and historian servers sit between IT and controls. Operating systems go out of support, certificates expire, backups are never tested, and the first sign of trouble is a server that will not restart.

Historians that record the wrong things. Thousands of tags collected at default rates because licensing allowed it, while the few signals that would explain a quality problem were never configured. A historian is excellent at continuous process values; it is not an OEE system, and buying it as one disappoints.

Polling that misrepresents the process. A supervisory system that polls every few seconds cannot see an event that lasts a few scans. We explain why in PLC scan cycles and SCADA polling.

The decisions this layer forces

Which platform. Our position: for most discrete and hybrid plants starting fresh or replacing a fragmented system, Ignition is a strong default because its server licensing model does not charge per client or per tag, and it runs on standard databases the plant’s IT group can support. A plant already deep in FactoryTalk View SE or AVEVA, with people trained on it, may be better served by fixing its existing system than by switching. We will tell you which applies.

SCADA or HMI. Many machines only need a local HMI. A server-based system brings real infrastructure cost, covered in our page on FactoryTalk View SE vs ME.

What to historize. Decide the questions first, then the tags, rates and retention that answer them. Collecting everything and querying nothing is the common outcome when this step is skipped. Our complete guide to data historians covers those decisions by tag class.

Standards first. Tag naming, screen templates and an alarm philosophy belong in a written standard before the first screen is built, so the tenth line looks like the first.

How it connects to the rest of the stack

This layer reads from controls and from edge and I/O devices, over the OT network. It feeds plant data for reporting and analytics and supplies context to MES and batch. When the tag structure here is clean, everything above it gets easier; when it is not, every layer above inherits the problem.

Where Joltek helps

We start with a plant systems assessment and a modernization roadmap that decide what to keep, consolidate or replace. For new systems we act as owner’s engineer or support systems integration directly, and where the goal is downtime visibility we connect this layer to OEE and downtime improvement. On Rockwell sites, see our page on FactoryTalk Historian.

Platform position

Vendor-independent, but not opinion-neutral. When one platform is clearly better for a given situation, we will say which and why. When the honest answer is that the existing system is fine and the money is better spent elsewhere, that is the recommendation.