Stop polling remote sites

The polling tax

Classic polling asks every device for every value on a schedule, whether anything changed or not. Over a metered cellular or radio link, that means:

• Constant round trips and data charges for values that rarely move.
• Latency — you learn about an event on the next scan, not when it happens.
• Transients that slip between scans entirely.

Report-by-exception with Sparkplug B

Sparkplug B puts a lightweight edge node at the site that publishes values only when they change, with store-and-forward so nothing is lost when the link drops. Birth and death certificates tell the system exactly what tags exist and when a node goes offline. Ignition's MQTT Engine subscribes, and the tags simply appear in your tag tree.

What changes on the gateway

• No scan classes pounding a slow link — the edge decides what's worth sending.
• Events arrive in near real time instead of on the next poll.
• Adding a site is publishing to a broker, not wiring up another polled connection.

For water districts and anything on cellular, this is usually a dramatic cut in both data cost and lag.