The Data Volume Problem
A single PLC variable sampled at 1 Hz generates 31.5 million data points per year. With 50 variables across 10 machines, that is 15.75 billion rows annually. Storing raw data forever is impossible. The solution is tiered aggregation.
Tiered Aggregation Strategy
Raw data (1-second resolution): keep for 30 days. 1-minute aggregates (min/max/avg): keep for 1 year. 1-hour aggregates: keep for 5 years. Daily aggregates: keep forever. This reduces storage by 99.7% while preserving the ability to answer almost any historical query.
TimescaleDB for Time-Series Data
TimescaleDB is a PostgreSQL extension that handles time-series data 10–100x faster than standard MySQL for append-heavy workloads. Its built-in time_bucket() function makes aggregation queries simple. It supports automatic data retention policies — raw data deletes itself after 30 days.
Cost of 10 Years of Machine History
With tiered aggregation, 10 years of data for a 50-machine factory fits in approximately 50–100 GB. On AWS RDS, this costs under ₹3,000/month. The insight value of historical data — identifying seasonal patterns, predicting failures, proving compliance — is worth far more.
Never delete machine data — aggregate it instead. A well-designed tiered retention policy keeps 10 years of history at under ₹3,000/month in cloud storage costs.
Need Help With This?
Want to start logging your machine data properly? We design time-series data architectures for Indian manufacturers.
Talk to Our Team →