SCADA, Telemetry & Industrial Monitoring challenge
Twelve questions on historians, deadbands, time sync and the OT/IT boundary.
SCADA / OT · Interactive Operations Model
Historian Storage Calculator
Capacity · Growth · Retention
Change the archiving assumptions and see which of them your storage estimate actually rests on — tag count, sample interval, record size, retention or the compression you expect to achieve. Everything downstream is that rate multiplied by a decision.
Points actually written to the archive, which is usually far fewer than the points configured in SCADA.
Interval at which the source produces samples, before the historian compresses anything. Use the scan or acquisition rate — the reduction from a deadband belongs in the compression ratio below, and counting it here as well would apply it twice.
How much history stays online. Older data is usually moved to slower storage rather than deleted.
Effective reduction the historian achieves: deadband or swinging-door compression together with any file compression. Steady analogue data reaches 10:1 and beyond; alarms and fast digitals barely compress at all.
Start from an example: Mid-size plant, one-second data, one year online →
Advanced parameters1
- —Archived tags
- —Sampled every
- —Uncompressed / day
- —Compression
- —Kept for
- —Storage
- —Storage required
- —Growth per day
- —Growth per year
- —Before compression
- —Sample rate
Move a parameter and the archive answers: what arrives every second, what is left after compression, and what that comes to over the retention asked for.
The link carries the figures you entered, in the part of the address that is never sent to a server.
Change one assumption at a time. The model updates from the same calculation used for the reported result.
Try this
Keep the tag count and retention fixed. Change only the sample interval from 5 seconds toward sub-second sampling and watch which part of the storage requirement moves fastest.
How the model works
- The rateEvery archived tag produces one sample per interval. The tag count and the interval together set the sample rate, and everything downstream is that rate multiplied by something.
- The recordEach sample is written as a record — timestamp, value, quality. Bytes per sample is the width of that record before compression touches it.
- The squeezeHistorians compress hard, because process data repeats. The ratio is the one number here that is assumed rather than counted, and it divides the whole answer.
- The yearsThe archive grows at a constant rate, so retention is simply a multiplier — which is why the honest question is usually how long, not how much.
Need sizing based on the real SCADA configuration?
We can model the actual tags, update rates, retention and replication — and turn the sizing into something your team reruns as the plant changes, instead of a spreadsheet nobody trusts.
Calculation basis 4 steps · view the method →
Samples per second = archived tags ÷ sample interval—Uncompressed per day = samples per second × bytes per sample × 86,400 s—Uncompressed total = uncompressed per second × retention period—Storage required = uncompressed total ÷ compression ratio—
Assumptions & limitations Screening estimate for capacity planning and budgeting · 7 assumptions →
This is a screening estimate. Real archives depend on the compression actually achieved by your data, on indexes, replication and backups, and on how the product stores its files. Confirm capacity with the vendor's sizing guidance and with measurements from a running archive before committing hardware.
- One average sample interval is applied to every tag; in practice fast and slow tags differ by orders of magnitude.
- The sample interval is the rate before compression; the deadband's effect belongs entirely in the compression ratio and must not be counted in both.
- Storage is expressed in decimal units, as disks are sold: 1 TB = 10¹² bytes.
- The compression ratio is applied uniformly across all tags.
- Indexes, replication, snapshots, backups and filesystem overhead are excluded.
- Event, alarm and batch records are excluded — these are usually small next to time-series data, but not always.
- The archive is assumed to run at a steady state, with no backfill or reprocessing.
Worked examples
- Mid-size plant, one-second data, one year online
Open with these figures → - High-rate subset, 100 ms data, 90 days online
Open with these figures → - Wide but slow: 60,000 tags at one minute, five years
Open with these figures →
What this model reveals
An archive is not sized by how much data a plant has, but by a rate multiplied by a decision. Tag count and sample interval set what arrives every second; retention multiplies it; compression divides the whole of it — and compression is the one number here that is assumed rather than counted. The model keeps those apart so you can see which of them your estimate actually rests on.
Go deeper
Power & Utilities expertise
Cloud SCADA, distributed data acquisition and digital twins around real infrastructure.
GridModel case study
A pilot-ready enterprise platform for the electric-utility industry.