GoogleCloudPlatform/gke-autoscaling-benchmarking

GKE Autoscaling Benchmarking

Purpose

Runs a series of benchmarks against a workload with HPA enabled within a GKE cluster.

Prereqs

• Ensure you can build docker images locally, without sudo (https://docs.docker.com/engine/install/linux-postinstall/#manage-docker-as-a-non-root-user, go/installdocker#sudoless-docker).
• Configure Artifact Registry and install the gcloud CLI credential helper (http://cloud.google.com/artifact-registry/docs/docker/authentication, go/installdocker#gcr-credential-helper)
• Ensure you're logged in via gcloud (Hint: gcloud auth login)
• Ensure you can run terraform
  • Install Terraform version 1.7.5 or higher with these instructions

Instructions

1. Manually create a project with a valid billing account. (These scripts will not do this for you).

2. Create the cluster via:

   cd stage02-cluster
   cp sample-terraform.tfvars terraform.tfvars
   vim terraform.tfvars  # edit the values appropriately
   terraform init
   terraform apply
   cd ..

3. Create the workload via:

   cd stage03-workload
   terraform init
   terraform apply
   cd ..

   This will create a deployment that will calculate numbers in the fibonacci sequence. Feel free to replace this with your own workload.

4. Create the benchmarking tooling (which will run on the same cluster):

   cd stage04-benchmarking
   terraform init
   terraform apply
   cd ..

5. Start the benchmarking via the locust web UI

   # Get the external IP Addr
   gcloud container clusters get-credentials benchmarking --location=us-central1-c
   kubectl get svc/locust -n fib -o json | jq '.status.loadBalancer.ingress[0].ip'
   
   # Now visit the ip addr at port 8089 in your web browser

   Run the test with 50 users to start.