IMD SolutionComputing LibraryShop main store

Explainer

Why Geo SCADA history needs points to backfill Citect trends

Edge computing and IIoT Published 6 min read

Short answer

No — not without duplicate points. Geo SCADA keeps history only for points with the Historic feature enabled, and the documented write path runs through that point model, for example the LoadDataValue method of a point's Historic aggregate. A Citect 2018 backfill therefore means creating the destination points, or choosing a SQL database or a Citect-integrated historian instead.

On this page
  1. What Geo SCADA stores history under
  2. Why the .NET API does not remove the point requirement
  3. Size the destination before you commit
  4. Four paths and when each one wins
  5. Running a backfill that survives unattended operation
  6. When the trend comes back empty
  7. The decision, case by case
  8. Questions asked

No, not tagless. Geo SCADA cannot act as a standalone historian fed by Citect SCADA 2018, because its history belongs to configured points: every stored sample lives under a point with the Historic feature enabled, and the documented write path — the LoadDataValue method of a point's Historic aggregate — writes into a point that already exists. Your scheduled transfer is buildable; the "no duplicate tags" half of the plan is not. Accept one destination point per Citect tag, or move the store to a SQL database or a historian with documented Citect integration. What follows is what the point model requires, the server numbers at backfill scale, and the four paths that work.

What Geo SCADA stores history under #

In Geo SCADA, history is a property of a point, not a separate database you pour records into. You enable it per point on the Historic tab of that point's configuration form; the same tab holds a Compression section where the criteria for historic processing and storage are defined. Schneider Electric's point configuration guide states the condition plainly: to keep a historical record of a point's values, you enable the Historic feature on that point.

That per-point model reaches into server sizing. Historic points are database points with the feature enabled, and the vendor's hardware guidance puts the typical point-to-object ratio at about 1:2, so every destination point you add for the backfill also grows the object database. The same guidance discusses systems of 50,000 points, so scale is not what stops this design — identity is.

Why the .NET API does not remove the point requirement #

The documented programmatic route for entering history is the LoadDataValue method on a point's Historic aggregate, and the vendor's community documentation describes using it to enter history to any point in the database. Read that back: any point. The method addresses an existing point object; it does not accept a free-floating tag name that Citect could map onto by itself. Reading and editing history through the API behave the same way — every operation names a point. Connecting an external database to Geo SCADA adds a data source; it does not add storage identity.

Rule out a second shortcut at the same time: moving the complete Citect database is not a larger version of moving history. Project objects and timestamped samples are different things, and a backfill moves only the samples.

Size the destination before you commit #

Worked example, real numbers. Citect SCADA 2018 is logging 1,200 trends you want retained two years online, with three or more operator clients.

  • Destination points required: 1,200 — one per tag; the model permits nothing less.
  • Objects created, at the documented 1:2 point-to-object ratio: 1,200 × 2 = 2,400 objects.
  • Documented server tier: 10,000 objects with 5,000 historic points, stored 2 years online, 3 or more clients — 4 cores, 8 GB RAM, 25 GB free space on high-performance disks.

2,400 objects is 2,400 ÷ 10,000 = 24% of that tier's object count, so take the 4-core, 8 GB, 25 GB machine as your floor even at this modest tag count: the transfer writer and the operator clients share it. At ten times the tags you reach 12,000 points — beyond the 5,000-historic-point tier — so re-size from the published tiers before committing.

Four paths and when each one wins #

Path What you must build Choose it when
Geo SCADA points plus .NET history writes One point per Citect tag with Historic enabled, plus a scheduled transfer service Geo SCADA is the mandated store and duplicate points are acceptable
Historian with documented Citect integration That product's tag configuration, through its documented interfaces Operators must keep native Citect trends
Separate SQL database Tables, tag mapping, scheduled writes, a query-backed display Duplicate SCADA points are unacceptable and you own the query layer
Replace Citect with Geo SCADA Full point, display and alarm rebuild in Geo SCADA A SCADA platform migration is already planned; this is not a historian project

Once the points exist, the first path is an ordinary backfill, and the discipline in backfilling trend history from a plant historian carries straight over: bounded ranges, read-back verification, progress persisted only after success. For the second path, weigh candidates the way plant historian selection by scale and support needs does — tag count, retention, and who supports it on nights. The third path's write side looks like logging production data to SQL Server: a timestamp and quality column, an index on time, scheduled inserts.

Retrieval is a separate task from storage, and it follows the rules of writing timestamped history back into a SCADA server: a successful import changes nothing about which source a Citect trend reads.

Running a backfill that survives unattended operation #

  1. Prove one tag end to end first: one bounded range — a single shift — extracted, written, read back, compared.
  2. Extract through Citect's supported archive interface and confirm every record carries timestamp, value and quality; you need all three downstream.
  3. Create the destination point and enable Historic on its Historic tab; set the compression criteria there before the first import, not after.
  4. Record the mapping for every tag: source trend, destination point, value type, scaling, and whether timestamps are local time or UTC — decide once, in writing.
  5. Import with the original sample times. The API's PresetTimestamp and PresetQuality fields exist for this. An imported record stamped with the transfer time instead of the sample time is false data carrying a good flag, so treat timestamp fidelity as a hard acceptance test.
  6. Read back and reconcile record counts and spot values before any display is connected.
  7. Schedule transfers and persist progress only after verified writes. Before unattended running, establish how the API treats a write against an already-stored timestamp, and make the retry resume from the last reconciled range.
  8. Wire retrieval last: point the query-backed display, or the integrated historian's trend, at the destination and test identity and time-range handling with a known range.

When the trend comes back empty #

Work from the display backwards before blaming the controller:

  • Identity and time range in the display: a wrong tag or window hides correctly stored samples.
  • Which source the trend actually reads: local archive or external destination.
  • Source availability in Citect: separates records that never existed from records lost in transfer.
  • The mapping table: samples written under the wrong point read as gaps everywhere else.
  • Point existence and Historic enablement in Geo SCADA: the silent failure mode of this design.
  • Live acquisition and driver status: only once neither store shows new samples.

Records present in Geo SCADA but blank in Citect mean retrieval or display binding. Present in Citect but missing in Geo SCADA mean extraction, mapping, permissions or writer errors. Absent in both: check source logging first, then acquisition.

The decision, case by case #

  • Geo SCADA is mandated as the store: accept the duplicate points — 1,200 in the worked example — and write through the documented history method with original timestamps. No tagless alternative exists; stop searching for one.
  • No duplicate points, as a hard requirement: use a SQL database and own the schema and the display queries.
  • Operators must keep native Citect trends: choose a historian with documented Citect display integration first; any other path makes the display side the largest part of the project.
  • A platform migration is already planned: replace Citect with Geo SCADA outright and run one system, rather than building a shadow historian for a plant about to migrate.

Questions asked

Does every backfilled point also need a live controller connection in Geo SCADA?

No. The Historic feature is a storage property of the point, separate from acquisition, and the documented examples of entering history cover internal points as well as I/O points. For a receive-only destination, build the mapping on internal points so nothing in Geo SCADA polls the controller; confirm the point class behavior in the documentation for your installed version.

What server size covers a 5,000-tag backfill, and what happens at longer retention?

The published tier for 10,000 objects with 5,000 historic points held two years online and three or more clients is 4 cores, 8 GB RAM and 25 GB free space on high-performance disks. The 25 GB figure is for two years online: stored volume tracks retained sample count, so four years of the same tags is roughly double the space, while cores and RAM stay put.

What happens if a scheduled batch fails halfway and retries?

Some records may already be committed when the batch reports failure, so a blind rerun depends entirely on how the API treats a write against an existing timestamp. Find that out deliberately: write one duplicate timestamp in a test view, observe the result, then design the retry to resume from the last reconciled range rather than the last attempted one.

Is moving the complete Citect database the same as backfilling its history?

No. A database move carries project objects — drivers, graphics, alarms, variable tags — while a history backfill carries only timestamped samples. They are different kinds of data with different tools, and moving the database does not copy the trend archive into a new store. Treat them as separate projects: migrate the project when you migrate the platform, backfill the samples when you migrate the history.

Do operators have to leave Citect to see the imported history?

Not necessarily, but it depends on which destination you choose. Native Citect trends read Citect's own archive, so samples written to Geo SCADA or to a SQL database will not appear in them. Keep operators in Citect by choosing a historian with documented Citect display integration, or build a query-backed display against the SQL destination and place it beside the Citect screens.

Keep reading

  • How-to

    Backfilling trend history from a plant historian

    Yes, you can backfill trend history from a plant historian into a SCADA server. Extract the stored records through the historian's native export or ODBC interface, map each source point to its destination object with original timestamps and quality, insert the batches through the destination's documented history methods, and prove the import by reading the interval back from the destination. A saved CSV is a bridge, not proof of import.

    Edge computing and IIoT

  • Comparison

    Operator access to historian trends on night shift

    Put routine historian trend access in front of operators: prebuilt HMI or SCADA trend screens for the recurring questions, point selection over already-collected points for everything else, and a read-only web portal as the alternate route. Keep collection configuration with engineering. No permission setting retrieves samples for a point the historian never collected, so that gap is closed by adding collection, not by access rights.

    Edge computing and IIoT

  • Explainer

    Do you need a historian or just an OPC tag server?

    No. An OPC tag server (OPC DA) supplies the current value, quality and timestamp of each item and stores nothing, so monthly figures need a historian or a SQL logger collecting from that same server. Keep a reporting client like ActiveFactory only if you also run the historian it expects.

    Edge computing and IIoT