Autoscaling and Initial Scaling

Michael Delzer Updated by Michael Delzer

When the User deploys the stack template to make a stack instance, the User is given options to define the initial number of worker nodes. When the User configures an application template, default starting replicas for microservices can be configured

There are 3 components of autoscaling: HPA (which is enabled in kube-apiserver and monitors pods CPU/memory requests), cluster-auto scaler component (which monitors unscheduled/pending pods from HPA), and metrics-server component(kubernetes add-on which provides HPA with CPU/memory metrics).

Different hosting options/cloud vendors have different features for auto scaling horizontal growth in the number of worker nodes.

In AWS adding worker codes is done on a deployed Stack. The AWS account being used must have rights and limits set to allow auto scaling worker pools to grow to the size the business is expecting.

To add an auto scaling pool to an AWS cluster (stack) find the Stack in the List view

Click on "Kubernetes" to change the view to the capacity view instead of the default "Components" view

Add a name, suggest reference stacassociateded with the pool, this makes the node names start with this name and becomes easier to support.

to let the cluster from from zero to a fixed number slide "auto scale" to On (slide to right)

Choose the instance type that meets the needs of horizontal scaling workload.

Choose spot price (AWS) or on Demand (Azure and GCP have different terms for reduced price options)

Choose how much storage the noods need.

Like what you see? Request a demo today!

How did we do?

Introducing AgileStacks to Your Cluster

Importing an upstream Kubernetes cluster