EC2 Launch Templates

An EC2 Launch Template is a configuration template that defines the parameters required to launch an EC2 instance. Introduced in 2017, it supersedes the older launch configurations (which are now legacy and not recommended for new deployments). The primary purpose of a launch template is to provide a reusable, versioned, and shareable blueprint for instance launches. This ensures consistency across multiple instances, simplifies management, and enables advanced features like Auto Scaling, Spot Instances, and fleet management.

When you create a launch template, you specify a set of parameters that AWS stores as a resource in your account. Each template can have multiple versions (starting at version 1). When you launch an instance using a launch template, AWS reads the template parameters and applies them to the new instance. The launch template is not a running entity; it is a static configuration object. The actual instance launch process follows these steps:

A launch template can include the following parameters (with defaults where applicable):

Versioning: Each launch template has a version number (integer, starting at 1). You can create new versions with updated parameters. The default version is the one you mark as default. When launching, you can specify: - $Default (the default version) - $Latest (the highest-numbered version) - A specific version number (e.g., 2)

Limits:

Using the AWS CLI, you can create a launch template with the following command:

aws ec2 create-launch-template \
    --launch-template-name MyTemplate \
    --version-description "Production version" \
    --launch-template-data '{
        "ImageId": "ami-0abcdef1234567890",
        "InstanceType": "t3.micro",
        "KeyName": "my-key-pair",
        "SecurityGroups": ["sg-0123456789abcdef0"],
        "UserData": "IyEvYmluL2Jhc2gKZWNobyAiSGVsbG8i"
    }'

To view existing templates:

aws ec2 describe-launch-templates

To view versions of a template:

aws ec2 describe-launch-template-versions \
    --launch-template-name MyTemplate

To launch an instance using a template:

aws ec2 run-instances \
    --launch-template LaunchTemplateName=MyTemplate,Version=$Default \
    --subnet-id subnet-0123456789abcdef0

Important: Launch templates do not store launch history or instance IDs. They are purely configuration objects. To see which instances were launched from a template, you must tag instances with the template name or version during launch.

A large e-commerce company runs a fleet of 100 web servers behind an Application Load Balancer. They use an Auto Scaling group with a launch template that specifies the latest Amazon Linux 2 AMI, t3.medium instance type, a security group allowing HTTP/HTTPS, an IAM role for S3 access, and user data that installs the web server and pulls the application code from S3. The launch template ensures every new instance is identical, reducing configuration drift. When a new AMI is released, they create a new template version and update the ASG to use $Latest. They gradually terminate old instances (using a lifecycle hook) to perform a rolling update. Without launch templates, they would need to manually specify all parameters each time they replace an instance, risking misconfiguration.

A data analytics company runs large-scale batch processing jobs on Spot Instances. They use a Spot Fleet request with a launch template that defines multiple instance types (e.g., c5.xlarge, c5.2xlarge, m5.large) and a common AMI and user data. The launch template allows them to specify the same configuration across different instance types. They also set instance metadata options to require IMDSv2 for security. The Spot Fleet automatically selects the cheapest available instance type from the list. If they need to update the AMI, they create a new version of the launch template and update the Spot Fleet request. This approach simplifies management and reduces costs by leveraging Spot Instances.

Launch templates themselves have no performance impact; they are configuration objects. However, using templates with many parameters (e.g., many block device mappings) can increase the API call size but not launch time. The number of versions does not affect performance. For large fleets, ensure your CloudTrail logging does not get overwhelmed by launch events.