OPC UA ETL Glossary
Short definitions of the terms used in OPC UA ETL. Each one links to the section of the complete OPC UA ETL guide where it is explained in context.
DAG (directed acyclic graph)
A flow diagram of a pipeline in which data moves in one direction, from sources through transforms to sinks, and never loops back.
Drawing an ETL pipeline as a DAG makes the whole flow visible at a glance and doubles as documentation. In OPC Data Wrangler you build the DAG on a visual canvas by connecting nodes. See a typical OPC UA ETL pipeline.
ELT (extract, load, transform)
A variation of ETL where raw data is loaded into the destination first and transformed there afterwards.
ELT is common in data warehouses, where the destination is powerful enough to transform data after loading. For OPC UA, the consumer is usually a SCADA or another OPC UA client that expects clean tags on arrival, so the transform step normally happens before the load. That is the ETL order described in the OPC UA ETL guide.
ETL (extract, transform, load)
The process of reading data from a source, transforming it into the form you need, and writing it to a destination.
In industrial settings the three steps are: extract values from OPC UA servers, transform them (units, formulas, validation, custom logic) and load the result to a destination. Read the full OPC UA ETL guide.
Formula evaluation
A transform that computes a new value from one or more inputs using a mathematical expression.
Typical uses are scaling, ratios and compensation, such as temperature-compensating a gas flow reading. It sits between simple unit conversion and full custom code. See Transform: units, formulas and Python.
OPC UA
A secure, cross-platform standard for exchanging industrial data between devices, servers and applications.
OPC UA standardises how data moves between systems. It does not decide which units or tag names your consuming systems expect, which is why a transformation step is still needed. See Why OPC UA needs ETL.
OPC UA broker
Software that sits between OPC UA data sources and consumers, collecting data from sources and republishing it in a consistent form.
A broker decouples what the sources publish from what the consumers expect. An ETL pipeline is what a broker runs inside. Read what an OPC UA broker does for more.
OPC UA ETL
Extracting data from OPC UA servers, transforming it into the units, names and shapes other systems expect, and loading it to a destination.
OPC UA gets the data moving; ETL makes it usable. The OPC UA ETL guide covers the topic end to end, and The OPC UA ETL Problem Nobody Talks About walks through the problem with screenshots.
Python transform
A transform step where you write the logic in Python, for cases that go beyond unit conversion and formulas.
Used for smoothing noisy sensors, range checks and outlier detection. A moving average is a typical example, shown in the transform section of the guide.
Sink
The pipeline node that writes transformed values to a destination.
In OPC Data Wrangler the main sink is an OPC UA server, so SCADA and other OPC UA clients can read clean tags without knowing the original sources. CSV export is also available. See Load: writing to OPC UA servers.
Source
The pipeline node that reads values from an upstream system, such as an OPC UA server.
A source connects to a server, lets you browse the address space and select the tags to read. See Extract: reading from OPC UA servers.
Tag
A named data point, such as a temperature or pressure reading, that a server exposes and a client can read.
In OPC UA, tags are nodes in the server's address space. ETL maps source tags to destination tags. See tag normalization.
Tag normalization
Making tags from different devices and vendors consistent in name, unit and structure so downstream systems can treat them alike.
For example, one PLC publishes Temperature_Tank1 in Celsius while the SCADA wants Plant1.Temp.Tank01 in Fahrenheit. Normalization is the combination of renaming, unit conversion and validation that bridges that gap. See Why OPC UA needs ETL.
Transform
The pipeline step that changes values between reading and writing: converting, calculating, smoothing, renaming or validating.
Most transformations fall into three levels: unit conversion, formula evaluation and custom logic in Python. See the guide.
Unit conversion
Converting a value from one engineering unit to another, such as Bar to PSI or Celsius to Fahrenheit.
The math is simple but the risk is high, because manual conversions are easy to get wrong. OPC Data Wrangler includes a built-in Unit Conversion node covering 11 industrial unit categories with over 50 units. See Level 1: unit conversion.
Want the full picture? Read the OPC UA ETL guide or the walkthrough The OPC UA ETL Problem Nobody Talks About.