Glossary
In industrial data acquisition the same terms mean different things on the automation side and on the IT side, and that shows in a procurement meeting straight away. This page collects the concepts that recur in DataPortia™'s documentation and quotations. The terms are grouped by topic, because that is how they are needed — in groups rather than in alphabetical order.
Short answer
OPC UA is the data transfer standard of industrial automation, a time-series database stores measurement values with their timestamps, and SOE means sequence-of-events analysis at millisecond resolution. These three explain most of what DataPortia does.
OPC UA and data acquisition
These terms concern how data moves from the automation system to the reporting system. All of DataPortia's data acquisition is based on the OPC UA protocol, so the concepts in this group come up in every deployment. How collection is implemented is described on the page OPC UA data acquisition.
- OPC UA
- Open Platform Communications Unified Architecture. A vendor-independent standard by which automation systems and the software running on top of them transfer data. The successor to the old, Windows-bound OPC Classic.
- OPC UA server
- The part of the automation system that makes measurement values and alarms available for reading. DataPortia does not include a server; it connects to the one the system already has.
- OPC UA client
- Software that reads from a server. DataPortia acts as a client and writes nothing to the server.
- Data Access (DA)
- The part of OPC UA that covers the reading of measurement values: temperatures, pressures, flows, status information.
- Alarms & Conditions (Part 9)
- The part of OPC UA that covers alarms and events. It brings with it the severity, the area and the original timestamp.
- Measurement point (tag)
- A single collected value on the OPC UA server, for example the reading of one sensor. A plant typically has hundreds or thousands of them; which ones are collected is decided during deployment.
- Subscription
- An agreement with the server about which points are sent to the client and how often. A subscription is more efficient than repeated polling, because the server sends only the values that have changed.
- Historical archive
- The OPC UA server's own store of past values. If it is enabled, DataPortia can import history retrospectively; if not, collection starts from the moment of installation.
- OPC Classic (DA/HDA)
- The old generation of OPC, tied to Windows COM technology. DataPortia does not support it directly — an OPC UA server or a wrapper is needed in between.
Storage and calculation
These terms concern where the collected data goes and how it turns into figures that can be reported. The storage arrangement determines how far back data can be queried and what that costs in disk space. The details are on the page Technical specifications.
- Time-series database
- A database optimised for timestamped measurement data: many writes, queries by time range. An ordinary relational database slows down quickly under this kind of load.
- TimescaleDB
- A PostgreSQL extension that turns it into a time-series database. Because it is an extension rather than a different product, the database is backed up and maintained with ordinary PostgreSQL tools.
- Compression
- Packing older data into less space on disk. In DataPortia compression starts automatically after two days and saves over 90 % of the disk space. Compression does not restrict querying.
- Retention period
- How far back data is kept. Adjustable, 24 months by default. This is a different thing from compression: the retention period determines what exists, compression only how much space it takes.
- Calculation point
- A formula-based derived measurement point calculated from measured values — power into energy (kW → kWh), for example. A calculation point adapts to the report's time range, so the same definition produces both the daily and the monthly figure.
- Throughput
- How many values per second the system takes in and writes to the database. In DataPortia over 2 000 values per second, roughly 172 million rows per day.
Alarms and events
These terms concern working out afterwards what went wrong. The decisive concept is the resolution of the timestamp: it determines whether the order of events can be established at all. The subject is covered on the page Alarms and events.
- Alarm
- A state that stays on for as long as the deviation lasts and that can be acknowledged. For example “temperature above limit”.
- Event
- A momentary record that something happened. An event is not acknowledged, because it has no duration.
- Timestamp
- The instant the automation system assigns to an event. DataPortia keeps it at millisecond resolution without truncating it — no resolution is added, but none is lost either.
- SOE (sequence of events)
- Sequence-of-events analysis. It tells you what tripped first and what followed from it. It requires millisecond-accurate timestamps: at one-second resolution dozens of events land on the same instant and the order cannot be established.
- Severity
- The numeric weight OPC UA assigns to an alarm. Used for filtering, so that an investigation can be narrowed to the alarms that matter.
- Pareto analysis
- Ranking causes in descending order of their total impact. It shows that the most frequently repeated alarm is not the same as the one that costs the most time.
Integrations
These terms concern moving data out of DataPortia into other systems. All traffic is outbound and read-only: nothing is written to the automation system through the interface. The implementation is described on the page REST API and webhooks.
- REST API
- An interface from which the receiving system fetches data with HTTP requests. DataPortia's version 1 is read-only and contains no write operations.
- API key
- An identifier that distinguishes the calling system. It also carries the rate limit, so it is worth giving each integration a key of its own.
- Rate limit
- 120 requests per minute per key. The limit is per key, so requests from different systems do not eat into each other's budget.
- Webhook
- The interface in reverse: DataPortia pushes an event to the recipient the moment it occurs, instead of the recipient asking for it.
- HMAC signature
- A digest computed with a shared secret, by which the recipient verifies that a webhook came from the right place and has not been altered in transit.
Availability and redundancy
These terms concern preparing for something breaking. In DataPortia that preparation sits at three separate levels, and each of them is configured and tested separately. The whole picture is on the page Redundancy and high availability.
- Redundancy
- A standby arrangement that takes over when the primary part stops working. Not the same thing as a backup: redundancy keeps the system running, a backup restores it afterwards.
- Redundant server pair
- Two OPC UA server addresses behind the same connection. If the primary does not respond, the connection moves to the other one automatically.
- Streaming replication
- PostgreSQL's own method, in which the transaction log is shipped as a continuous stream to a second database server. TimescaleDB requires physical replication specifically.
- Failover
- The switch from the primary to the standby system. It can be automatic or done by hand.
- Active-passive
- A configuration in which only one server does the work at a time. The standby server shows the user interface, but its data acquisition and alarm subscriptions are switched off until it takes over.
- DPFC
- DataPortia Failover Controller. Atorcom's own separate program, which monitors the PostgreSQL pair and performs the failover automatically — on Windows too, unlike pg_auto_failover and Patroni.
- Witness node
- A third machine, holding no data, that settles which database node is the primary. Without it no automatic failover is performed, but the databases themselves work normally.
General terms
These terms recur throughout the documentation and do not belong to any one area. The first of them is also DataPortia's most central product characteristic: everything runs on the customer's own server.
- On-premises
- The software runs on the customer's own server rather than in the supplier's cloud service. In DataPortia this applies to AI analysis as well: process data does not leave the plant's network.
- Local language model
- An AI model that runs on your own server. DataPortia uses Ollama, through which you can run a Qwen, Gemma, DeepSeek or Llama model, for instance, without an internet connection.
- Dashboard
- A view that gathers the chosen measurements onto one page in real time. In DataPortia an eight-column grid on which widgets are placed freely.
- Trend
- A chart showing how a measurement develops over time. Historical trends can be zoomed and saved as presets for reviews that repeat.
- Floating licence
- A licence form that is not tied to the hardware identifier of one machine. The practical reason for choosing it is most often redundancy, where a hardware-bound licence would be needed separately for each of the two servers.
Is a term missing?
The glossary covers the concepts that recur around DataPortia. If a quotation, a document or this site uses a term that is not here and whose meaning is not obvious, ask — I will add it. That is quicker than guessing, and a misunderstood term costs more in a procurement than asking does.