Skip to main content
Datadog infrastructure plans include a number of containers per host. Containers above that are billed as overage, at on-demand rates. nOps finds organizations paying on-demand container overage every day and estimates the saving from staying inside the allotment.

Why it matters

Sidecars, init containers and short-lived jobs count as containers, so container counts often grow faster than the number of hosts.

How nOps finds it

nOps reads the last 14 complete days of Datadog billing for each organization and product. An organization is flagged when it paid on-demand container charges on all 14 days. Included allotments are never billed, so every billed line is overage. The saving is the lowest daily on-demand charge times 30, which is the overage that was always there, and must be at least $5 a month. Overage already prepaid in your commitment is not counted.

Before you start

Access to the Datadog Agent configuration (for example the Helm chart or DaemonSet) on your hosts and clusters.

How to implement

  1. In Datadog, open Plan & Usage and the container views to see which containers count most.
  2. Exclude short-lived, sidecar and system containers from the Agent with container exclusion settings (for example DD_CONTAINER_EXCLUDE or the containerExclude value in the Helm chart).
  3. Where it makes sense, consolidate many small containers onto fewer hosts.
  4. Check container usage over the next weeks to confirm it is back inside the allotment.

Risks and considerations

  • Excluded containers are no longer monitored, so you lose their metrics. Exclude only containers you do not need to watch.
  • How far the count can be cut is your call, so the saving is an upper bound.
  • To roll back, remove the exclusion.