Using SimKube AMIs

How to locate, run and use SimKube AMIs.

AWS IAM Permissions

The basic AWS IAM permissions required to run SimKube AMIs.

  "Effect": "Allow",
  "Action": [
    "ec2:DescribeImages",
    "ec2:DescribeInstances",
    "ec2:RunInstances",
  ],
  "Resource": "*"

• If you plan to import or export traces in AWS S3 you will need permissions for those resources in S3.

{
  "Effect": "Allow",
  "Action": [
    "s3:PutObject",
    "s3:GetObject"
  ],
  "Resource": "arn:aws:s3:::<bucket-name>/*"
}

We have included a full IAM policy example in our reference section.

Locating the AMIs

SimKube AMIs are published to the AWS Marketplace and versioned.

You can find the latest AMIs by:

• searching for AMIs owned by ACRL
• using the AWS CLI to filter by name and region

aws ec2 describe-images \
  --owners 174155008850 \
  --filters "Name=name,Values=simkube-*" \
  --query "Images[].{
    ImageId: ImageId,
    Name: Name,
    CreationDate: CreationDate
  }" \
  --region us-west-2 \
  --output table

Launching the AMI

You can launch an EC2 instance using the SimKube AMI via:

• the AWS console
• the AWS CLI
• infrastructure as code (IAC) tools like Terraform / Pulumi

Here are the AWS docs on launching EC2 instances.

When launching an instance:

• select an appropriate instance type for your use
• configure your network and security groups
• provide an SSH key pair

Instance sizing

SimKube simulations are a compute bound workload.

Our recommended default instance type is c7a.2xlarge:

• 8 vCPUs
• 16 GiB RAM
• Strong price/performance for compute intensive workloads
• Cost efficient AMD architecture

This instance size supports most simulations reliably but we encourage you to experiment to find the right instance size for your specific simulation needs.

Accessing the instance

Instances launched from the SimKube AMI support SSH access.

• use the default ubuntu user
• authenticate using the SSH key pair specified at launch

AMI versioning and updates

Each AMI is versioned and immutable.

• any updates will be delivered by publishing a new AMI
• existing instances are not modified
• we recommend frequently updating for the latest bug fixes, improvements and base OS security updates

Limitations

• The AMIs are designed for and only available in AWS EC2
• long-running simulation environments should be managed and monitored by the user

Next steps

• Configure GitHub Actions to run simulations on self-hosted runners