Observability Best Practices¶

Guidelines for getting reliable signal from Observability while keeping the noise (and the maintenance) low. For guidance on configuring the checks themselves, see the best practices for Volumetric, Freshness, and Metric checks.

Turn on tracking early¶

Volume Tracking and Freshness Tracking record measurements from the moment you turn them on, and history cannot be filled in later. Turn them on for every container that matters to your data quality, not just the ones you plan to check today: the history is what change comparisons compare against, and what AI Managed checks learn from. An AI Managed Freshness check needs at least eight days of history, and AI Managed Volumetric checks need from about a week (the 1 day check) to about sixteen weeks (the 28 day check).

Pair every chart with a scheduled Scan¶

The charts update on their own, but they never raise an anomaly. A salmon point or bar on a chart means the value is outside the limits and no Scan evaluated the check. Schedule a Scan with the Metadata category for containers with Volumetric or Freshness checks, and with the Data Integrity category for Metric checks, so a value outside the limits becomes an alert.

Read the heatmap, not just the anomaly count¶

The heatmap surfaces context the anomaly count misses:

A run of cyan squares confirms the container was measured every day.
A grey square in a cyan row means no measurement was recorded that day, not that nothing happened. Change comparisons are not asserted on Scans whose period includes it.
An orange square marks a period in which the selected check raised an anomaly. Hover for the counts.

Use the heatmap for triage before opening individual anomalies.

Watch for Inaccessible containers¶

A container whose measurements fail three times in a row is marked Inaccessible and stops being measured, without retrying. Its heatmap stays grey until someone fixes access and runs a Sync, or, for a computed container, edits, validates, and saves its definition. Review the status on the cards you depend on, especially after credential or network changes.

Favorite the containers you review daily¶

The Bookmark icon on a card marks it as a favorite and pins it to the top of the listing. Favorites are per user, so each team member can pin the containers they own. Combined with tag filters, favorites turn a large observability listing into a personal dashboard.

Use the chart editors for limits only¶

The inline Edit Thresholds and Edit Maximum Age editors on the chart change only Min and Max, or the Maximum Age. Everything else on the check (Description, Tags, Metadata, comparison, field) is edited from the check form. On an AI Managed check, an edit from the chart converts it to Authored, the same as an edit from the form.