Kedify at KubeCon 2026: Explore autoscaling and rightsizing with our team
Kedify turns live workload signals into right-sizing recommendations, autoscaling action, fleet coordination, and cost evidence.
Collect demand, pod pressure, scaling history, and capacity across clusters.
Telemetry intake
Kedify combines workload demand, pod pressure, scaling history, and fleet capacity so decisions are based on live behavior, not CPU alone.
Right-size CPU and memory requests with confidence before scaling acts.
Insights
Insights surfaces CPU and memory recommendations with confidence, explanations, and safe commands before resource changes are applied.
Autoscale APIs, queues, jobs, GPU inference, predictions, and pod resources.
Autoscaling action
Kedify acts on HTTP traffic, queue pressure, predictive demand, GPU workloads, and vertical pod resources from the same control loop.
Coordinate policies across clusters with guardrails, weights, and failover.
Fleet control
Multi-cluster scaling uses advanced patterns to place work across member clusters with weights and failover.
Tie saved pod-hours, node-hours, CPU, memory, and GPU to FinOps impact.
FinOps
FinOps estimates saved pod-hours, node-hours, CPU, memory, and GPU capacity against recent peaks so savings are visible and explainable.
Kedify helps teams cut idle spend and protect responsiveness with autoscaling that reacts to real demand, right-sizes resources, and coordinates across clusters.
Scale capacity from real demand, right-size CPU and memory, and turn saved capacity into FinOps evidence.
HTTP traffic, queues, events, custom metrics, and GPU utilization drive capacity instead of static overprovisioning.
Tune replicas, CPU, and memory together so workloads keep enough headroom without carrying unused requests.
Show saved pod-hours, node-hours, CPU, memory, and GPU capacity in FinOps-ready views.
Keep services responsive when traffic spikes, queues build up, or GPU inference demand changes.
Scale APIs, queues, jobs, and custom metric workloads from live signals before users feel the bottleneck.
Use predictive scaling and GPU-aware policies to place capacity ahead of recurring demand.
Apply multi-cluster weights, failover, and guardrails across teams, tenants, and environments.
Most teams validate Kedify in one cluster first, using real metrics to confirm scaling behavior, right-sizing recommendations, and cost impact.
14 days: a paid proof of concept in 1 cluster, one scaling path, with a $5k credited fee. Applied to the annual contract when you move forward.
| Plan | Price | Clusters | Scaling |
|---|---|---|---|
| Starter | $800/mo + usage (billed annually) | Up to 3 | 70+ event-driven scalers, Insights, and FinOps visibility |
| Professional | $2,000/mo + usage (billed annually) | Up to 10 | Starter capabilities + HTTP/gRPC, predictive, and vertical scaling |
| Enterprise | Speak with Sales | Unlimited | Professional capabilities + multi-cluster, GPU/AI, and custom scaling patterns |
– Surag Mungekar, CISO, Rupert
AWS • GCP • Azure • OpenShift • Kubernetes
Prometheus • OpenTelemetry • 75+ event sources (Kafka, Redis, RabbitMQ, and many more )
