Connectors
Connectors for plant, historian and business data
Aditum Data reads OPC UA, MQTT (including Sparkplug B) and Modbus TCP through a small edge connector at the site, reads AVEVA PI and GE Proficy historians, and connects to the databases, warehouses and streams you already run. Data goes to the destination you choose: your own platform, or the Aditum database with live dashboards.
This page lists what is available in production today, how each connection is made, and what it can and cannot do.
Facts checked
Which systems can Aditum Data connect to?
Plant protocols and historians as sources; databases, warehouses, data lakes and streams as sources and destinations.
| System | Direction | How it connects |
|---|---|---|
| OPC UA (including Kepware and ABB 800xA through their OPC UA servers) | Source | Edge connector at the site (Windows or Linux) |
| MQTT, with Sparkplug B decoding | Source | Edge connector at the site |
| MQTT publish (for example a unified namespace) | Destination | From Aditum, plain JSON, QoS 1 |
| Modbus TCP | Source | Edge connector at the site |
| AVEVA PI System (PI Web API) | Source, read-only | From Aditum, or the on-site reader inside the plant network |
| GE Proficy Historian | Source and destination | From Aditum |
| PostgreSQL and TimescaleDB, MySQL | Source and destination | From Aditum, or the on-site reader (as a source) |
| Microsoft SQL Server, Oracle | Source | From Aditum, or the on-site reader |
| MongoDB | Source and destination | From Aditum |
| InfluxDB | Source and destination | From Aditum, or an InfluxDB at the site |
| Snowflake, Databricks, Google BigQuery, Amazon Redshift | Source and destination | From Aditum |
| Amazon S3, Azure Data Lake and Blob Storage (including Delta tables) | Source and destination | From Aditum |
| Apache Kafka, Azure Event Hubs | Source and destination | From Aditum |
| SAP (OData V2 and V4) | Source, read-only | From Aditum |
| REST APIs, Salesforce | Source and destination | From Aditum |
| SharePoint and OneDrive (documents) | Source | From Aditum |
| Files: CSV, Excel, JSON, text | Source; CSV and Excel export | Upload in the web app |
| The Aditum database (hosted PostgreSQL with TimescaleDB) | Destination | Hosted by Aditum Data in the EU |
"From Aditum" means the connection is made from Aditum's servers in Frankfurt; those connections reach only public internet addresses. A system inside the plant network is read through the edge connector or the on-site reader instead.
OPC UA
The edge connector reads OPC UA servers inside the plant and sends the values out over HTTPS on port 443. Nothing opens an inbound port, and nothing is written back to the server.
- Browse and search the server's whole address space in the web app, then pick the tags to read.
- The server certificate is verified before reading.
- After a network outage, gaps are filled from the server's own history (OPC UA HistoryRead) where the server keeps history.
- Kepware KEPServerEX and ABB 800xA are read through their OPC UA servers.
- Tag selection, deadband, downsampling and filters are set centrally in Aditum, versioned and can be rolled back, with no site visit.
MQTT and Sparkplug B
The edge connector subscribes to your MQTT broker and decodes Sparkplug B payloads; Aditum can also publish prepared data back to an MQTT broker, for example into a unified namespace.
Published messages are plain JSON with QoS 1, so any subscriber can read them without an Aditum library.
Modbus TCP
The edge connector polls Modbus TCP devices on the plant network and sends the readings out the same way as OPC UA: outbound HTTPS only, buffered during outages.
OPC DA (OPC Classic)
Today, OPC DA servers are read through an OPC UA gateway such as Kepware KEPServerEX, which the OPC UA connector then reads.
Historians: AVEVA PI System and GE Proficy
Aditum reads your existing historian rather than replacing it. Read the history once, then only what is new.
AVEVA PI System
Aditum reads PI points through the PI Web API, read-only. When the PI Web API is only reachable inside the plant network, the on-site reader makes the connection from inside and sends data out over HTTPS. PI points can also be placed on the plant graph (site, unit, equipment).
GE Proficy Historian
Aditum reads from and writes to GE Proficy Historian.
Databases and time-series stores
PostgreSQL (with TimescaleDB), MySQL, MongoDB and InfluxDB work as sources and destinations; Microsoft SQL Server and Oracle as sources.
- A connection test for PostgreSQL, MySQL, SQL Server, Oracle and InfluxDB lists each step (DNS, TCP, TLS, sign-in, first query) and says in plain words how to fix the one that failed.
- Kepware DataLogger tables are recognised when Aditum reads the database they are logged to.
- Power BI, Excel or Qlik can read the Aditum database through a read-only SQL login.
- Scripts and lab systems can write time series into the Aditum database through the time-series write API.
Warehouses, data lakes and streams
Prepared plant data can land in Snowflake, Databricks, Google BigQuery, Amazon Redshift, Amazon S3 or Azure Data Lake (including Delta tables), or on Kafka and Azure Event Hubs.
Any table Aditum can read can also be exported as Parquet, CSV or JSON Lines, and scheduled exports to S3-compatible storage are available through the API.
SAP, REST APIs and documents
SAP is read through OData (V2 and V4), read-only; REST APIs and Salesforce work in both directions; SharePoint and OneDrive supply documents such as P&IDs and datasheets.
SAP RFC and BAPI are not used.
How the edge connector works
One small read-only service per site, on Windows or Linux, that sends data out over HTTPS on port 443 and buffers it when the connection drops.
- Runs as a Windows service under a low-privilege account, or as a systemd service under a non-login Linux user. The Windows package installs offline.
- Outbound HTTPS on port 443 only, optionally through your company's HTTP, HTTPS or SOCKS5 proxy. No inbound port into the plant.
- Read-only towards the plant: the connector contains no code that writes to a control system.
- Store and forward: during a network or Aditum outage, data is buffered in memory and on disk (with a size cap) and replayed in order when the connection is back.
- One install can read several sources at a site, with one process, buffer and key per source.
- Window values keep the real minimum, maximum and count, so a spike is not lost; averages are time-weighted.
- In local mode, OPC UA, MQTT and Modbus data can be written to an InfluxDB at the site, and only health information goes to Aditum. That data then does not reach Aditum's dashboards.
- An edge connector shows as offline 3 minutes after its last heartbeat, and an alert opens within ten minutes.
- Edge connectors and streaming are part of the Pro plan and above.
Questions
Does Aditum Data write anything back to the PLC or DCS?
No. The edge connectors are read-only towards the plant and contain no code that writes to a control system.
What does IT have to open in the firewall?
Outbound HTTPS on port 443 from the machine that runs the edge connector, optionally through your proxy. No inbound port is needed.
What happens to the data when the network goes down?
The edge connector buffers it in memory and on disk and replays it in order when the connection is back. For OPC UA, gaps can also be filled from the server's own history.
Do we have to replace our PI System or historian?
No. Aditum reads AVEVA PI and GE Proficy as sources and delivers prepared data to the platforms you already run.
Can we connect OPC DA servers?
Yes, through an OPC UA gateway such as Kepware KEPServerEX, which the OPC UA connector reads.
Discuss architecture with an industrial data specialist.
Source compatibility, security boundaries, deployment and multi-site operation. Or request a 45-minute technical session to select one KPI and three sites and draft the acceptance criteria.