Back to all posts
Plant Data7 min read·October 25, 2026

SCADA Logging vs Historian: What Does Your Plant Need?

TB
ThynkBlox Automation Team
Automation Engineering

SCADA logging is sufficient when it can retain the records you need and retrieve them reliably for the people who use them. Consider a separate historian when historical analysis, longer retention or access across systems exceeds what your existing setup can demonstrate. For SCADA logging vs historian, make the decision around actual queries and data behaviour, then include the work of maintaining another system.

What questions must your historical data answer?

Start with a short list of operational questions, not a tag count.

An operator might need the temperature trend before the last trip. A maintenance lead might need to compare the same machine before and after a repair. A process engineer might need to compare batches made under different conditions.

Turn each question into a retrieval test:

  • Which signals and events must appear together?
  • How far back must the search reach?
  • Must the result show individual recorded values or a summary?
  • What response time is acceptable?
  • Who needs access, and through which screen or tool?

A trend that works for one shift does not establish whether the same setup meets your requirements for older records, simultaneous users or batch comparisons. Demonstrate those cases separately.

When is SCADA logging sufficient?

Keep the existing arrangement if it passes the retrieval tests, meets retention requirements and has a clear support owner.

Do not assume that adding a historian will improve an investigation. First identify the specific limitation: missing records, slow queries, difficult exports, disconnected systems or insufficient context.

Decision areaKeep SCADA logging if it demonstrates…Evaluate a historian if you need…
Recent investigationsReliable access to the required trends and eventsHistorical queries that the current setup cannot serve acceptably
RetentionAccessible records for the agreed retention periodA separate long-term archive with a demonstrated retrieval method
Cross-system analysisA workable way to align the relevant recordsShared access to history from several systems
Process contextRecords linked to the equipment, product or batch being investigatedA defined way to associate process history with contextual records
Operational ownershipSupported backup, recovery and administrationA separately owned data service with its own support arrangements

Treat the right-hand column as reasons to evaluate an option. A product label is not evidence that the proposed system meets those needs.

What should you test in historical queries?

Ask the supplier or integrator to demonstrate the questions your team actually asks.

Use a recent operating period, an older period and a period containing an interruption. Include a query across several signals and a request made while normal collection continues.

Specify what each returned value means. Is it an original recorded sample, an average, a minimum, a maximum or an estimated value between samples? Require the display and export to make that distinction clear.

Also define how queries handle gaps, duplicate timestamps, uncertain readings and changes to tag names or units. Ask whether timestamps represent measurement time or receipt time, and how the system handles clock differences.

These requirements belong in the functional design specification. Agree the expected result before accepting a demonstration.

How should retention and compression be assessed?

Separate the retention policy from the storage estimate.

Decide how long each class of record must remain available, at what detail and for which purpose. A recent fault investigation and a long-term production comparison may need different detail. Do not apply one retention setting to every signal without reviewing those uses.

Ask whether the proposed compression is lossless or whether it reduces the samples retained according to configured rules. Where samples are reduced, require a clear explanation of what may be omitted and how that affects retrieval.

For changing process values, test steady operation, gradual movement, a brief excursion and an abrupt change. For discrete states and counters, define which transitions or increments must remain visible.

Check collection settings separately. If a short event is absent from the incoming data, changing archive settings cannot supply the missing observation.

Reject a storage proposal that offers only an estimated disk size. Require the assumptions behind it and evidence that the retained detail still answers your operational questions.

What context must accompany a trend?

Define the records needed to interpret each investigation: equipment identity, product, batch or order, operating mode, alarm state and relevant maintenance activity.

For every contextual record, name its source and owner. Specify how it is linked to the process history, how corrections are recorded and what happens when it arrives late.

Do not assume a historian includes production tracking or maintenance records. Ask where those records will live and how a user will retrieve them alongside a trend.

For example, if the question concerns an OEE change, agree which production counts, planned periods and downtime reasons must accompany the process values. Our guide to what the OEE number hides provides a useful starting point for those questions.

What does a practical comparison look like?

Illustrative example: A plant wants to investigate a temperature excursion during a batch. It needs to compare that batch with an earlier batch of the same product and identify the operating mode during each excursion.

Give both proposed arrangements the same acceptance test:

  1. Retrieve both batches using their identifiers.
  2. Display temperature, setpoint and operating mode over each batch period.
  3. Identify which points are recorded samples and which are summaries or estimates.
  4. Show any missing or uncertain data clearly.
  5. Export the values with timestamps, units and batch identity.
  6. Repeat the retrieval after restoring an archived backup in a test environment.

If the existing SCADA arrangement passes within the agreed response time, the test has not established a need for a separate historian.

If it fails, record the exact failure. Evaluate whether a configuration change, a contextual-record interface or a historian addresses that failure. Avoid buying a larger system to solve an undefined problem.

Who will maintain a separate historian?

Include operational ownership in the purchase decision. Request a support plan covering:

  • Collection failures, connection loss and recovery of queued data.
  • Storage growth, retention changes and archive availability.
  • Tag additions, renamed equipment and changed engineering units.
  • User access, software updates and supported interfaces.
  • Backup, restoration and periodic retrieval checks.
  • Ownership of reports and downstream connections.

For each item, name the responsible party and the evidence that the task has succeeded.

Ask what happens when the historian is unavailable and whether collection resumes without manual intervention. Define how missing periods are reported and how recovered records are reconciled.

Include licences, implementation, training and ongoing administration in the comparison. A system without an agreed support owner is an incomplete proposal.

What common mistakes should you avoid?

Choosing by tag count alone. Use the actual collection behaviour, retention policy and query workload in the evaluation.

Accepting a recent trend as the whole demonstration. Test older archives, interruptions, exports and concurrent requests.

Treating a smooth line as proof of complete records. Require visibility of recorded points, estimated values and gaps.

Ignoring context. Make batch, equipment and operating-mode associations part of acceptance.

Using process history as a safety function. Keep this article's data-storage decision separate from safety-related design, which should be addressed with a qualified assessor.

Related reading: setting historian deadbands without hiding events, diagnosing gaps in historian data, setting historian data retention rules and giving production teams access to historian data.

What are the frequently asked questions?

Can we keep SCADA and add a historian later?

Make future access a design requirement now. Confirm export rights, available interfaces, timestamp conventions and ownership of historical records. Do not assume that a future migration will preserve all detail without a demonstrated method.

Does longer retention automatically require a historian?

No. Require the existing arrangement to demonstrate the retention period, recovery process and historical queries you need. Decide from those results.

Should every tag be archived?

Give each archived signal a purpose. Review what it supports, the detail required and its retention period. Include relevant states and events alongside analogue values.

What should we request before approving a project?

Request a written retrieval test, retention policy, compression explanation, context model and support plan. Use our plant data services to discuss the scope, or contact us with the questions your current records cannot answer.

Next step

Want to apply this to your plant?

Every project starts with a fixed-fee assessment, and you can stop after the report.