SCADA, an industrial historian and a general time-series database may all display trends, but they solve different problems. The architecture should assign acquisition, operational context, durable history, analytics access and record ownership explicitly instead of treating the products as interchangeable stores.
The short answer
Keep real-time supervision and operator interaction in the control-system boundary. Use a historian when dependable industrial collection, compression, quality, event context and long retention are required. Use a general time-series platform where approved analytics or application workloads benefit from flexible development and scale. Decide with requirements for latency, loss tolerance, timestamp authority, quality semantics, backfill, retention, query load, redundancy, cybersecurity and ownership. One system can serve several roles, but every role still needs explicit acceptance tests.
- Separate control, operational history and analytical consumption responsibilities.
- Preserve source timestamps, quality and operating context.
- Design for store-and-forward, gaps, clock errors and recovery.
- Document authoritative records and consumer limitations.
Decision guide
| Platform role | Best fit | Main design risk |
|---|---|---|
| SCADA | Supervision, alarms, operator displays and near-real-time control context | Overloading the control environment with broad analytical demand |
| Historian | Durable industrial history with collection and quality semantics | Poor tag/context governance or untested recovery |
| Time-series database | Application and analytics workloads with flexible development | Assuming generic ingestion preserves industrial meaning |
| Archive/export | Regulated or contractual evidence packages | Confusing a snapshot with the operational source of truth |
A practical workflow
- 01
Define consumers, decisions, latency, availability, retention and recovery objectives.
- 02
Map sources, clocks, quality semantics, rates, events and network boundaries.
- 03
Assign each platform role and identify the authoritative copy for each use.
- 04
Test normal collection, overload, outage, backfill, failover and query isolation.
- 05
Publish ownership, interfaces, limits, backups, recovery and change controls.
Evidence and acceptance
Retain architecture and data-flow diagrams, tag and context ownership, protocol and interface versions, timestamp and quality mapping, rate and retention calculations, compression rules, cybersecurity zones, identity and access model, outage/backfill tests, redundancy evidence, backup and restore results, performance tests, consumer contracts and handover responsibilities.
Common mistakes
- Sending every high-frequency value to every platform.
- Dropping quality and source timestamps during integration.
- Allowing analytics queries to affect operational performance.
- Assuming replication is a tested backup.
- Leaving tag naming, retention and decommissioning ownerless.
Responsibility boundary
Primary sources
Technical claims and product boundaries were checked against these primary sources. Fact-checked
Frequently asked questions
Can SCADA replace a historian?
Some products include historical functions, but suitability depends on retention, recovery, quality, scale and governance requirements.
Is a time-series database safe for control?
A general database must not be assumed to provide the deterministic or safety functions of the control system.
Where should timestamps originate?
Define an authoritative source and clock architecture, then test mapping and drift across the complete path.
Does AutoNexit store live process history?
No. It stores controlled project records and evidence, not historian telemetry.