How does PI data compression work?

PI Data Archive uses a swinging door compression algorithm to store only meaningful data changes, dramatically reducing storage while preserving data fidelity.

How Swinging Door Works

1. When a new value arrives at the snapshot, PI compares it to the last archived value.
2. It draws imaginary lines ("doors") from the last archived point, angled by the compression deviation (CompDev).
3. If all intermediate points fit within the doors, they're discarded.
4. When a point falls outside the doors, the previous point is archived, and the doors reset.

Key Tag Attributes

Attribute	Purpose
CompDev	Compression deviation — how much a value must change to be archived
CompMin	Minimum time between archived events (prevents flooding)
CompMax	Maximum time between archived events (ensures periodic archiving)
ExcDev	Exception deviation — filters noise at the interface level
ExcMin	Minimum time between exceptions sent to PI
ExcMax	Maximum time between exceptions

Compression vs Exception

• Exception happens at the interface/data source — filters before data reaches PI.
• Compression happens at the PI Data Archive — filters what gets permanently stored.
• Together they form a two-stage noise reduction pipeline.

Best Practices

• Set CompDev to 1-2% of the tag's typical range.
• Always set CompMax (e.g., 8 hours) so you see periodic values even for stable signals.
• Use ExcDev at the interface to reduce network traffic.
• Don't set CompDev to 0 unless you truly need every value (e.g., event-triggered tags).