Skip to main content

​
Act on Recommendations

KOSTRA continuously evaluates your connected accounts against a catalog of optimization and security checks, and records dated, evidence-backed findings. This guide shows how to review those recommendations and turn them into savings.

​
Open Recommendations

  1. In the left navigation, open Recommendations.
  2. Review the summary of potential savings and the list of recommendation types.
Each recommendation type — such as abandoned instances, unattached volumes, obsolete snapshots, idle Elastic IPs, abandoned OBS buckets, or database rightsizing — groups the individual resources it has flagged, along with the estimated monthly saving for each.

​
Filter to what matters

Two filters make a large estate manageable:
  • Category — separate cost-saving recommendations from security findings so you can route them to the right team.
  • Applicable services — narrow to a specific service on any provider. For Huawei Cloud, for example, you can filter to ECS, ELB, RDS, EVS, EIP, or OBS, so a team responsible for those workloads sees only its own opportunities.
Filtering by service is the fastest way for a team to find its highest-value actions without wading through the whole multi-cloud list, whichever provider it works on.

​
Review a recommendation

  1. Select a recommendation type to expand the resources it has flagged.
  2. For each resource, review the evidence and the estimated monthly saving. KOSTRA shows why a resource was flagged — for example, an OBS bucket flagged as abandoned because it is empty or its objects have not changed within the threshold window.
  3. Sort by saving to prioritize the biggest wins first.
Because every finding is dated and evidence-backed, you can trust the number and explain it to a resource owner rather than presenting an opaque score.

​
Take action

KOSTRA identifies the opportunity; you execute the change in the provider, following your own change process:
  1. Confirm the resource is genuinely a candidate — cross-check with the owner if needed.
  2. Make the change in the cloud provider (for example, delete an abandoned bucket, detach and remove an unused volume, or resize an over-provisioned database).
  3. On the next optimization run, KOSTRA re-evaluates and the resolved item drops off the list, so the recommendation list reflects reality over time.

​
Tune out the noise

Not every flagged resource should be acted on. KOSTRA lets you exclude resources — for example, by pool — so that intentional, known-good resources stop appearing as recommendations. Use exclusions deliberately so the list stays a trustworthy action queue rather than being ignored.

​
A practical workflow

A routine that works for any team, on any provider:
  1. Open Recommendations and filter Applicable services to the services your team owns.
  2. Sort by estimated saving.
  3. Work the top items with their owners.
  4. Re-check after the next run to confirm the savings landed.
The experience is identical across AWS, Azure, Google Cloud, and Huawei Cloud — the same routine applies no matter where a resource lives.