Scaling up to +10,000 pods on Amazon EKS - AWS UG Perú Conf 2023

Prerequisites

• Terraform
• AWS CLI
• Helm
• Kubectl

Deployed resources

1. Amazon AWS

• Virtual Private Cloud (VPC)
• Identity and Access Management (IAM)
• Elastic Container Service for Kubernetes (EKS)
• Amazon EKS managed node (EKS)

2. Kubernetes

• Metrics Server
• Karpenter
• App + Load

Variables

Name	Description	Type	Default	Required
PROJECT	Project's name	string	ugperu	no
ENV	Environment name	string	lab	no
AWS_DEFAULT_REGION	Amazon AWS Region	string	us-east-1	no
EKS_VERSION	Kubernetes version	string	1.27	no

Use

1. Create EKS cluster

make cluster

2. Configure CNI

make cni

3. Install components

make components

4. Install Karpenter

make karpenter

5. Install demo application

make app

6. Install loading app

make load

Debug

1. Observe pod scaling

kubectl get hpa --watch

2. Observe node scaling

kubectl get nodes --watch

3. Observe states of the ReplicaSet

kubectl get rs --watch

4. Observe the state of Karpenter

kubectl logs -f -n karpenter -l app.kubernetes.io/name=karpenter -c controller

Destroy

1. Stop pod scaling

make reset

2. Delete infrastructure

make destroy

Authors

The demo is maintained by Ivan Echegaray