Skip to main content

Labels & Tags

Labels (or tags) are used to attach context to metrics and events so that you can filter, aggregate, and route them effectively. A deliberate label strategy is essential for readable dashboards, actionable alerts, and scalable storage. This guide describes a practical labeling strategy for Unified Monitoring.

Start with a small, consistent set of labels:

  • Cloud scope – cloud, account_id / subscription_id / project_id, region, az.
  • Business context – application, service, environment, business_unit, owner.
  • Technical context – cluster, role, tier, component.
  • Compliance/risk – data_classification, criticality where applicable.

These labels should align with your CMDB attributes so searches, dashboards, and reports tell a coherent story.

Mapping from Cloud Tags​

Cloudaware typically derives labels from:

Maintain clear tag standards (naming and allowed values) and enforce them via governance or automation where possible.

Cardinality and Performance​

Avoid label designs that generate an unbounded number of unique combinations, such as:

  • Using user IDs, request IDs, or IP addresses as labels.
  • Embedding highly variable strings (for example, full URLs) as labels instead of fields in log data.

Symptoms of excessive cardinality include slow queries, high storage consumption, and dashboards that are difficult to interpret.

Use labels for dimensions you commonly filter on; store highly variable data in logs or event payloads instead.

Operational Tips​

  • Document your label taxonomy and share it with application teams.
  • Create validation rules or CI checks to enforce tag standards in infrastructure‑as‑code.
  • Use dashboards that highlight “unlabeled” or “mis‑labeled” resources as part of monitoring coverage reviews.