Skip to content

Rightsizing

Rightsizing finds instances and volumes that are larger than they need to be and tells you how much you would save by trimming them. It turns "we might be over-provisioned" into a ranked list of concrete, costed actions.

What it analyses

OpenWatch reviews two kinds of resources for your organisation:

  • Instances (Nova): flavors that look oversized for their observed usage, with a suggested smaller flavor.
  • Volumes (Cinder): disks that could move to a smaller size or a cheaper type.

It also surfaces snapshot chains that have grown long and are worth pruning.

Every analysed resource gets two numbers:

  • an efficiency score (higher is better) that reflects how well the resource is sized, and
  • an estimated monthly saving if you apply the recommendation.

Resources are grouped as downsizable, already optimal, or "no data" (not enough usage history to judge yet). The page also shows an average score and the total potential saving, with the biggest saving listed first.

How recommendations are produced

Rightsizing reads cached cost and usage data rather than querying your cloud on every visit, and the analysis runs on a schedule (roughly every six hours) in the background.

  • The first time you open the page, OpenWatch starts an analysis and shows a "loading" state. It fills in when the result is ready.
  • You can trigger a fresh analysis manually with the refresh action.

Score over time

The page tracks your efficiency score across recent months and draws it as a small trend chart. Use it to confirm that cleanup work is actually moving the needle, and to catch drift when new, oversized resources creep back in.

Estimates, not guarantees

Savings are estimates based on the pricing grid and observed usage. Treat them as a prioritised to-do list. The real saving depends on your provider's billing once the change is live.

Acting on a recommendation

You can act in two ways.

From your cloud console (always available)

Read the recommendation, then resize the instance or change the volume in your provider console. This keeps you fully in control and is the safest path when a workload is sensitive.

From OpenWatch (optional, applies to your cloud)

OpenWatch can also apply an instance resize for you through the Nova API, individually or as a batch, and tracks the action so you can follow it up.

A resize restarts and changes real infrastructure

Applying a resize acts on your live cloud, not just on OpenWatch. An instance resize typically reboots the server and, on OpenStack, must be confirmed once it completes. Plan it in a maintenance window, verify the target flavor, and make sure the workload tolerates a restart. When in doubt, apply the change from your console.

OpenWatch never resizes or retypes anything on its own. Nothing changes unless you explicitly trigger the action.

Availability

Rightsizing is a gated feature; availability depends on your plan. See Editions and plan limits.

Suggested workflow

  1. Open the page and let the analysis complete.
  2. Start with the largest estimated savings.
  3. For each item, confirm the workload can take a smaller flavor or disk.
  4. Apply the change (console or OpenWatch), preferably in a maintenance window.
  5. Watch the score trend over the next weeks to confirm the gain and catch regressions.