Skip to content

Getting Started

Use this page for the first hosted EcoScale workflow: connect a cluster, confirm that workloads have recommendations, and review one low-risk change.

Before you begin, confirm that you have:

  • An EcoScale account with access to Settings -> Clusters.
  • Kubernetes access to install the EcoScale agent Helm chart.
  • Metrics for the workloads you want to evaluate.
  • Access to view recommendations in EcoScale.

Install the agent with mode=readonly first. Use mode=apply only when the team is ready for EcoScale to patch target workloads. For installation options, RBAC, and secret handling, see Agent Helm Chart.

EcoScale needs cluster, workload, and metrics data before it can produce useful recommendations. If you want cost impact estimates, configure CPU and memory pricing before prioritizing work by projected savings.

For cluster setup, see Clusters. For recommendation fields and review criteria, see Recommendations.

Start with a cluster or namespace your team understands well.

  1. Confirm that the cluster appears in EcoScale.
  2. Open the workload inventory and select a workload with populated CPU or memory recommendations.
  3. Compare current requests and limits with the recommended values.
  4. Confirm the workload, namespace, cluster, and container names before taking action.
  5. If the recommendation is clear, apply it manually during an appropriate change window and validate the result.

A successful first rollout proves that the cluster is connected, recommendations are usable, and the team has a repeatable review path.

Hosted mode is the EcoScale cloud/SaaS distribution. Cluster and workload access are scoped by tenant and user context, which keeps multi-cluster and multi-team environments separated inside the platform.

AreaWhat to confirmWhy it matters
Cluster visibilityThe target cluster appears in EcoScale.Recommendations need the correct cluster context, especially in multi-cluster environments.
Workload discoveryWorkloads appear with target references.Operators must know exactly which Kubernetes object a recommendation affects.
Metrics historyRecommendations are populated for selected workloads.Resource guidance depends on observed behavior.
Cost configurationPricing is configured when savings estimates are needed.Financial prioritization depends on realistic CPU and memory pricing.
Change permissionsApply actions are available only to the right users.Resource updates mutate Kubernetes workloads and should be controlled.

If any of these checks fail, resolve visibility and data quality before applying recommendations or enabling automation.

Use Workload Optimization for workload discovery and prioritization. Use Recommendations for CPU and memory guidance. Use Automation only after manual review has produced predictable results.