RealTheory’s Optimization Profiles help tailor right-sizing recommendations based on the intent of your workloads. Not every workload should be treated the same — a critical customer-facing service has different performance needs than a development environment or a batch job.
Optimization Profiles enable you to prioritize what's most important: performance, cost savings, or a balance of both. These profiles influence the guidance RealTheory provides when recommending changes to CPU and memory requests and limits.
Important: RealTheory will always recommend cluster and workload configurations that meet baseline compute needs. The profile strategy simply influences the resource allocation changes while respecting performance needs.
Profiles can be applied manually or automatically through Optimization Policies, which allow you to assign profiles at scale across your organization (e.g., by cluster, namespace, or labels).
Choosing the Correct Profile
Choosing the correct Optimization Profile helps ensure RealTheory recommendations align with your operational goals. Each profile is best suited to a particular optimization strategy, and using the most appropriate profile will give you the most accurate recommendations based on the characteristics of your workloads:
| Profile | Description | Example Use Case |
|---|---|---|
| Performance Optimized | Recommends cluster and workload configurations that not only meet baseline compute demands but also include additional overhead to absorb unexpected surges, ensuring low latency and uninterrupted performance during peak loads. | A high-traffic API service that must respond quickly, even under variable load. |
| Balanced | Recommends cluster and workload configurations that strategically moderate resource allocation between performance and cost. | Workloads that experience regular demand with moderate peaks, such as a batch processing workload |
| Cost Optimized | Recommends the most economical cluster and workload configurations that satisfy compute demands while minimizing over-provisioning. | A dev or staging environment where resource efficiency matters more than performance. |
See Also
Setting the Default Optimization Profile
Creating an Optimization Profile