Data Backfill is a feature that stores temperature measurements locally on your sensor when it loses connection to the cloud. Once the connection is restored, the sensor automatically sends the stored data back to the cloud so you don't lose any readings. This article explains which sensors support it, how it works, and how long backfilling takes.
Who is this article for?
Anyone using DT temperature sensors who wants to understand how offline data is recovered, or who has noticed gaps in their temperature data after a connectivity interruption.
What do you need?
- A Temperature sensor (with Data Backfill enabled), a Temperature Probe sensor, or a Temperature & Humidity sensor (Data Backfill is included by default on Probe and Humidity sensors).
- A stable and permanent installation with good and reliable connectivity to a Cloud Connector.
Where should Data Backfill be used?
Data Backfill is primarily designed to prevent data loss during unexpected interruptions such as service outages or power failures. It works best in stable, permanent installations where connectivity is reliable most of the time.
While it can also handle planned disconnections or be used in data-logging scenarios, a stable connection must be re-established after any offline period. Consistent connectivity is recommended for optimal performance.
How Data Backfill works
When a sensor goes offline, it stores temperature measurements in its local memory. Once the connection to the cloud is restored, the sensor automatically begins sending the stored data, starting with the most recent samples first. If the sensor's memory becomes full while offline, the oldest data is overwritten.
The number of data points that can be stored during an offline period depends on the sampling rate, heartbeat configuration, and temperature fluctuations.
Storage capacity by sensor type:
- Temperature sensor - up to 100,000 events
- Temperature Probe sensor - up to 50,000 events
- Temperature & Humidity sensor - up to 31 days of data
The backfill duration is not affected by the sampling rate. Touching the sensor will not speed up the backfilling process.
How long does backfilling take?
After a sensor reconnects to a Cloud Connector, it takes at least two heartbeats before backfilling begins:
- The first heartbeat triggers the retrieval request.
- The second heartbeat is when the sensor responds and starts sending stored data.
Once backfilling starts, the last 48 hours of stored data are sent at an accelerated rate of up to 1 heartbeat interval (HBI) per second. After that initial burst, the rate slows to up to 20 HBIs per heartbeat.
Quick estimate: Minutes of offline time ÷ HBI = Time needed for backfill in seconds (after the 2nd heartbeat)
If all the data is not backfilled within the first 48 hours, the backfill rate will then slow to up to 20 backfilled HBIs for every 1 HBI.
Whether a reported temperature event came from backfill is currently only visible through the API. The event contains a field called isBackfilled, documented in the Developer Docs. This information is not yet available in Studio.
Troubleshooting / FAQ
Q: Can I use Data Backfill to intentionally take sensors offline for data logging?
A: While Data Backfill can handle planned disconnections, it is primarily designed for unexpected outages. You must re-establish a stable connection after offline periods for data to be recovered. Consistent connectivity is recommended for best results.
Q: Will touching or interacting with the sensor speed up backfilling?
A: No. The backfill duration is not affected by the sampling rate or by physically touching the sensor. The process runs automatically at a fixed rate once connectivity is restored.
Q: How can I tell if a temperature reading was backfilled?
A: Currently, this is only visible through the API. Each temperature event includes a field called isBackfilled. You can find details in the Developer Docs. This is not yet shown in Studio.
Q: What happens if the sensor's memory fills up while offline?
A: The sensor overwrites the oldest stored data first, so the most recent measurements are always preserved.