Kedify Features
Kedify features extend the upstream KEDA model with operational controls that are difficult to build and maintain in a large Kubernetes estate. These pages focus on the parts of autoscaling that sit around the core KEDA and HPA loop: tenancy, fleet operations, policy guardrails, vertical optimization, recommendations, and cost visibility.
What Kedify Features Add to KEDA
Section titled “What Kedify Features Add to KEDA”- optimize cost and performance with in-place vertical scaling, not only replica changes
- manage larger fleets with multi-cluster and multitenant control patterns
- review dashboard recommendations and estimated autoscaling savings
- apply autoscaling guardrails consistently across teams, namespaces, and clusters
- add workload lifecycle controls such as pausing, resuming, and resource-profile transitions
Feature Areas
Section titled “Feature Areas”- Pod Resource Profile: reusable CPU and memory profiles for pods.
- Pod Resource Autoscaler: automatic pod resource adjustments from demand.
- Multi-Cluster Scaling: autoscaling patterns across clusters.
- Multitenant KEDA: safer KEDA operations for shared environments.
- Insights: CPU and memory recommendations for workloads.
- FinOps: estimated spend, peak capacity, and savings.
- Scaling Policy: guardrails for consistent scaling behavior.
- Scaling Groups: shared scaling policy for related workloads.
- Resume Scaling Controller: pause and resume autoscaling configuration.
Recommended Starting Points
Section titled “Recommended Starting Points”- Compare Pod Resource Profiles and Pod Resource Autoscaler when horizontal scaling alone is not enough.
- Review Multi-Cluster Scaling and Multitenant KEDA if you operate more than one team, namespace, or cluster.
- Use Insights and FinOps when you want to connect utilization recommendations with estimated capacity and cost savings.
- Use Scaling Policy and Scaling Groups if you need safer default behavior across multiple workloads.
- Review Resume Scaling Controller when you need to pause autoscaling without removing the scaling configuration.