For modern cloud-native enterprises, achieving resource efficiency is a strategic necessity. Across fleets of clusters, infrastructure costs can increase, and performance predictability can suffer due to under-utilized CPU and memory allocations. Over-burdened workloads jeopardize compromising reliability and service-level objectives. To help platform engineers and operators address these challenges at scale, Red Hat Advanced Cluster Management for Kubernetes 2.16 introduces the general availability (GA) of right-sizing recommendations for namespaces and Red Hat OpenShift Virtualization workloads, delivering a unified, insightful, and observability-driven approach to resource optimization across multicluster environments. With this release, right-sizing recommendation moves from early innovation to fully supported enterprise capability.
From Tech Preview to GA
In our previous article, we discussed how right-sizing recommendation in Red Hat Advanced Cluster Management 2.14 began as a namespace-focused technology preview designed to help administrators understand the gap between actual CPU and memory usage and configured resource requests.
The integrated Grafana dashboards provide a feature that exposes the following sustained utilization trends:
- Over-provisioned namespaces with idle resources
- Under-provisioned namespaces at risk of throttling
- Opportunities to rebalance resource allocation safely
Building on that foundation, Red Hat Advanced Cluster Management 2.15 introduced right-sizing recommendation insights for OpenShift Virtualization workloads. This expansion extended optimization visibility from containerized applications to virtual machines (VMs), enabling administrators to evaluate CPU and memory overestimation or underestimation at the VM level.
Now, with Red Hat Advanced Cluster Management 2.16, namespace and virtualization right-sizing recommendation capabilities reach general availability, fully integrated, production-hardened, and supported for enterprise environments.
3 benefits of right-sizing recommendation GA
This section details the key deliverables of the right-sizing recommendation GA in Red Hat Advanced Cluster Management 2.16, focusing on its resource insight, data-driven methodology, and visualization through integrated Grafana dashboards.
1. End-to-end resource insight
Right-sizing recommendation now provides a consistent, fleet-wide view of resource allocation efficiency across clusters, namespaces, and virtual machines (OpenShift Virtualization).
By correlating long-term historical CPU and memory usage with requested allocations, Red Hat Advanced Cluster Management exposes inefficiencies that traditional point-in-time monitoring tools often miss. Administrators can evaluate real workload behavior instead of relying on static configuration assumptions.
2. Data-driven recommendations
Right-sizing recommendation in Red Hat Advanced Cluster Management is powered by the multicluster observability stack, including Prometheus recording rules and Thanos-based aggregation. This architecture enables:
- Trend-based analysis over sustained usage windows
- Aggregation across managed clusters
- Recommendation logic grounded in stable workload behavior
Because insights are derived from long-running time-series data, tuning decisions are based on evidence, not transient spikes.
3. Insightful visualization
The Red Hat Advanced Cluster Management embedded Grafana dashboards remain the primary interface for consuming right-sizing recommendation insights. The Namespace right-sizing recommendation dashboard compares requested versus actual CPU and memory usage across namespaces, highlighting optimization opportunities. The VM right-sizing recommendation dashboard evaluates per-VM resource allocation against historical consumption, surfacing over- and under-sizing patterns.
Interactive tables, utilization trends, and estimation variance views transform raw metrics into actionable intelligence. Whether you are optimizing microservices or virtualized legacy workloads, these dashboards ensure that measurable data drives resource decisions.
The following demo illustrates how to navigate the Grafana dashboards.
Why this matters to your organization
There are three primary organizational benefits of using right-sizing recommendation: cost savings, improved performance stability, and simplified multicluster management through unified insight.
Over-provisioned resources accumulate silently across clusters. By aligning CPU and memory allocations with actual usage patterns, organizations can significantly improve infrastructure utilization, especially in environments managing hundreds or thousands of namespaces and VMs thereby improving cost efficiency.
Under-provisioned workloads risk throttling, degraded performance, and instability. Right-sizing recommendation helps administrators proactively detect misalignment between allocation and demand, improving workload predictability and reliability.
Right-sizing recommendation operates within Red Hat Advanced Cluster Management’s multicluster observability framework, eliminating the need for disconnected optimization tools. Containerized and virtualized workloads are analyzed through a single management plane, ensuring consistent governance across environments.
Right-sizing recommendation vs. vertical pod autoscaler
Right-sizing recommendation and vertical pod autoscaler (VPA) optimize CPU and memory in Red Hat OpenShift, but they solve different problems and complement each other. The most effective approach combines Red Hat Advanced Cluster Management's right-sizing recommendation capabilities as the analytical layer with VPA serving as the execution and feedback layer.
Right-sizing recommendation provides an organization-wide view, surfacing which clusters, namespaces, or virtual machines are chronically over- or under-provisioned. It generates recommended CPU/memory baselines, which teams use to update their deployment manifests or policies.
VPA works within those recommended ranges, dynamically adjusting pod requests as workload traffic and behavior change. This ensures running workloads stay close to the optimal resource envelopes without requiring constant manual intervention.
Key differences at a glance:
Right Sizing Recommendations feature
- Goal: Fleet-wide efficiency and cost optimization
- Action: Recommendation only
- History: Long-term (Weekly/Monthly)
- Primary User: Platform Admin / FinOps / IT Management
- Reporting: Central dashboards across clusters, namespaces and virtual-machines
Vertical Pod Autoscaler feature
- Goal: Keeping individual pods correctly sized during their life
- Action: Recommendation and automation
- History: Short to mid-term (hours/days)
- Primary User: Application developers / SREs
- Reporting: Object-level recommendations close to workloads
Get started with right-sizing recommendation
To leverage the GA right-sizing recommendation capabilities, follow these steps:
- Upgrade to Red Hat Advanced Cluster Management 2.16.
- Enable multicluster observability, including Prometheus, Thanos, and Grafana components.
- Access the integrated right-sizing recommendation dashboards.
- Review recommendations across clusters, namespaces, and VMs.
- Gradually adjust namespace quotas or VM allocations based on sustained trends.
Because recommendation thresholds are configurable, administrators retain full control over optimization aggressiveness while maintaining workload stability. For more information and known limitations, refer to the official product documentation.
Looking ahead
The right-sizing recommendation GA release in Red Hat Advanced Cluster Management 2.16 represents a significant milestone. Try it and share your thoughts on the Red Hat OpenShift feedback form. Your input will shape the path to continued innovation and future enhancements. In future Red Hat Advanced Cluster Management releases, we plan to extend right-sizing recommendation dashboard support for Perses, further enhancing visualization flexibility and user experience.
