PLC & Data
Historian Storage Sizing
Storage a process historian will consume from tag count, sample rate, retention, and compression, with the daily and annual growth.
Calculated reference results
- Storage for the retention period
- 181.6 GB
- Per day
- 99.5 MB
- Per year
- 36.3 GB
- Raw data rate before compression
- 414.7 MB per day
- Samples per tag per day
- 17,280
- Total samples per day
- 34,560,000
- Sustained write rate
- 400samples per second
1,825 days
Averaged. Bursts during an upset are far higher.
The arithmetic
- Samples per tag per day = 86400 s / 5.00 s = 17,280
- Raw per day = 2,000 tags x 17,280 x 12 bytes = 414.7 MB
- After 5.0:1 compression and 20% overhead = 99.5 MB per day
- Over 1,825 days = 181.6 GB
What this does not account for
- Compression ratio is highly process dependent and is the least reliable input here. A steady analog compresses well; a noisy one barely compresses at all. Measure it on a pilot rather than assuming.
- Set compression to protect the excursions that matter. Aggressive deadbands erase the short events an investigation depends on, and that loss is permanent.
- Size for burst write rate, not the average. During an upset, alarm and event traffic can be orders of magnitude above steady state.
- Backup, replication, and any archive copies multiply this figure. Plan for the total footprint rather than the primary store alone.
How this is calculated
Historian storage is tag count multiplied by samples per day, multiplied by bytes per sample, multiplied by retention days. Compression, which stores a value only when it changes by more than a configured deadband, typically reduces this substantially, but setting it aggressively erases the short excursions that matter most in an investigation.
Formulas
bytes = tags x (86400 / interval) x bytes_per_sample x days / compression x (1 + overhead)
Assumptions and limitations built into this calculator
- Uniform sample interval across all tags, which is rarely true in practice.
- A single flat compression ratio, where real historians compress each tag differently.
- Does not include backups, replicas, or archive copies.
Frequently asked questions
- What compression ratio should I assume?
- Do not assume one. Historize a representative sample of real tags for a week and measure it. Ratios between 3:1 and 20:1 are all common depending on how noisy the process signals are.
Direct contact
Have a controls question?
Reach Eric Sullivan directly about anything on this site, a controls or automation topic, or one of his personal projects.