HPA
The Horizontal Pod Autoscaler in Kubernetes, which automatically adjusts the number of Pod replicas based on load metrics to achieve on-demand scaling.
Timeline (last 90 days)
HPA is Kubernetes' native horizontal Pod autoscaling mechanism, which periodically collects metrics (such as CPU, memory or custom metrics), compares them with target thresholds, and then proportionally adjusts the number of Pod replicas of controllers such as Deployment, ReplicaSet or StatefulSet.
HPA supports setting minReplicas and maxReplicas to bound scaling, avoiding unbounded growth or scaling to zero.
In Agent Substrate, the WorkerPool can be configured as the scaling target of HPA, scheduling pressure (the number of Actors queued waiting for memory) becomes the signal that triggers scaling, and the WorkerPool can scale up to 10 replicas.
HPA deals with increases and decreases in the number of horizontal replicas, whereas VPA (Vertical Pod Autoscaler) adjusts the resource quota of a single Pod.
This article chose HPA rather than VPA because the memory limit of a single Worker determines that it can only hold a fixed number of Actors; enlarging a single Worker's memory cannot solve the scheduling bottleneck, and increasing the number of Workers is the correct scaling direction.
Through the suspend-wake mechanism and HPA-driven Worker scaling, Agent Substrate shifts the scaling strategy from pre-allocating large amounts of resources to dynamically scaling according to actual scheduling demand, in order to reduce resource waste and guarantee response speed.
All Facts (6)
HPA is Kubernetes' native horizontal Pod autoscaling mechanism, which periodically collects metrics (such as CPU, memory or custom metrics), compares them with target thresholds, and then proportionally adjusts the number of Pod replicas of controllers such as Deployment, ReplicaSet or StatefulSet.
50%UnverifiedHPA supports setting minReplicas and maxReplicas to bound scaling, avoiding unbounded growth or scaling to zero.
50%UnverifiedIn Agent Substrate, the WorkerPool can be configured as the scaling target of HPA, scheduling pressure (the number of Actors queued waiting for memory) becomes the signal that triggers scaling, and the WorkerPool can scale up to 10 replicas.
50%UnverifiedHPA deals with increases and decreases in the number of horizontal replicas, whereas VPA (Vertical Pod Autoscaler) adjusts the resource quota of a single Pod.
50%UnverifiedThis article chose HPA rather than VPA because the memory limit of a single Worker determines that it can only hold a fixed number of Actors; enlarging a single Worker's memory cannot solve the scheduling bottleneck, and increasing the number of Workers is the correct scaling direction.
50%UnverifiedThrough the suspend-wake mechanism and HPA-driven Worker scaling, Agent Substrate shifts the scaling strategy from pre-allocating large amounts of resources to dynamically scaling according to actual scheduling demand, in order to reduce resource waste and guarantee response speed.
50%