Finding the Sweet Spot: Catching PI Tag Compression Issues with Osprey

PI exception and compression settings reduce unnecessary event traffic and archive storage while preserving useful signal behavior.

Poor settings can create two different problems. Aggressive filtering can hide important variation. Very loose filtering can store excessive events without adding useful information.

Understand exception and compression separately

Exception filtering is applied before events are sent to the PI Data Archive in many traditional interface architectures.

Compression determines which received events are stored in the archive.

The exact behavior depends on the collection architecture and point configuration. Review the complete data path before tuning settings.

Do not use one standard for every signal

A pressure transmitter, binary state, vibration signal, laboratory result, and calculated KPI have different data characteristics.

Set filtering based on:

• Engineering range

• Expected process variation

• Required fidelity

• Update frequency

• Downstream calculation needs

• Event and storage cost

Look for signs of over-compression

Potential indicators include:

• Stored trends that miss known process movement

• Step changes that appear later than expected

• Calculations that lose sensitivity because source variation is absent

• Similar assets with very different event density

Compare archived data with a higher-fidelity source or temporary test configuration before you change production settings.

Look for signs of under-compression

Potential indicators include:

• Very high archive event rates

• Repeated events with little new information

• Significant storage growth from low-value signals

• Unnecessary calculation or retrieval load

High event rate alone does not prove bad configuration. Some fast-changing signals legitimately require high fidelity.

Review related point metadata

Compression behavior depends on settings such as span and deviation values. Incorrect engineering range information can make percentage-based settings behave differently than intended.

Review engineering units and span together with filtering configuration.

Test before broad deployment

Use representative signals from each class. Compare stored trends, summary calculations, and downstream use before and after the proposed change.

Do not change thousands of tags only because a general rule appears reasonable.

Monitor after tuning

Track event rates, data-quality complaints, calculation behavior, and archive growth after a change.

Compression optimization is an engineering balance between fidelity and data volume. The correct setting preserves the process behavior that users and calculations need without storing unnecessary events.