Skip to content
Kedify Docs

For the complete documentation index and AI-optimized content, see /llms.txt. All pages support markdown format via .md extension or Accept: text/markdown header.

Kedify Architecture

For the complete documentation index and AI-optimized content, see /llms.txt. All pages support markdown format via .md extension or Accept: text/markdown header.

Kedify Architecture

Kedify extends the standard Kubernetes autoscaling stack instead of replacing it. KEDA still evaluates events and custom metrics. Kubernetes HPA applies replica changes by default; installations with KPA enabled can use the tenant-scoped Kedify Pod Autoscaler (KPA) for selected ScaledObjects. Kedify adds the operational pieces teams usually need in production: managed KEDA operations, HTTP autoscaling, richer observability, and centralized multi-cluster control.

How Kedify, KEDA, and the Horizontal Autoscaler Fit Together

Section titled “How Kedify, KEDA, and the Horizontal Autoscaler Fit Together”

At a high level, the control loop works like this:

  1. Your workload exposes demand through an event source, HTTP traffic pattern, or custom metric.
  2. KEDA or a Kedify scaler translates that demand into autoscaling metrics.
  3. Kubernetes HPA, or KPA when explicitly selected, uses those metrics to scale between one and many replicas. KEDA continues to own activation and scale to zero.
  4. Kedify provides the management, telemetry, policy, and advanced scaler capabilities around that loop.

This means Kedify is compatible with the core KEDA model while improving how teams operate autoscaling at scale.

Core Components

Section titled “Core Components”

  • Kedify Agent: The Kedify Agent is responsible for managing the installation and lifecycle of KEDA and the HTTP Scaler. It handles telemetry collection, providing users with valuable insights into their configurations. The agent establishes a secure, outbound gRPC connection to the Kedify Service, ensuring secure communication.

  • Kedify Dashboard: The Dashboard facilitates the management of installations and offers insights based on the telemetry data collected by the Kedify Agent. It serves as the central interface for monitoring and configuring Kedify-managed resources.

  • Kubernetes Integration: The Kedify Agent is deployed as a standard Kubernetes deployment. It manages Kedify custom resources, providing autowiring and comprehensive management of the HTTP Scaler. For more information on custom resources, refer to Kedify Custom Resource Definitions.

  • Telemetry Collection: The Kedify Agent includes a telemetry collector that scrapes metrics from both KEDA and the HTTP Scaler. These metrics are crucial for providing real-time insights and for monitoring the performance and scaling behaviors of your applications.

Where Kedify Extends Upstream KEDA

Section titled “Where Kedify Extends Upstream KEDA”

Kedify is most useful when you need more than raw event-driven autoscaling:

By combining these components, Kedify gives teams a more complete Kubernetes autoscaling architecture while retaining the native KEDA/HPA path and offering KPA where stronger tenant control-plane isolation is required.

For more detailed information about each component, please refer to their respective documentation: