AWS Service Catalog Construct Library

AWS CDK v1 has reached End-of-Support on 2023-06-01. This package is no longer being updated, and users should migrate to AWS CDK v2.

For more information on how to migrate, see the Migrating to AWS CDK v2 guide.

AWS Service Catalog enables organizations to create and manage catalogs of products for their end users that are approved for use on AWS.

Table Of Contents

• Portfolio

  • Granting access to a portfolio
  • Sharing a portfolio with another AWS account

• Product

  • Creating a product from a local asset
  • Creating a product from a stack
  • Creating a Product from a stack with a history of previous versions
  • Adding a product to a portfolio

• TagOptions

• Constraints

  • Tag update constraint
  • Notify on stack events
  • CloudFormation template parameters constraint
  • Set launch role
  • Deploy with StackSets

The @aws-cdk/aws-servicecatalog package contains resources that enable users to automate governance and management of their AWS resources at scale.

import aws_cdk.aws_servicecatalog as servicecatalog

Portfolio

AWS Service Catalog portfolios allow administrators to organize, manage, and distribute cloud resources for their end users. Using the CDK, a new portfolio can be created with the Portfolio construct:

servicecatalog.Portfolio(self, "Portfolio",
    display_name="MyPortfolio",
    provider_name="MyTeam"
)

You can also specify optional metadata properties such as description and messageLanguage to help better catalog and manage your portfolios.

servicecatalog.Portfolio(self, "Portfolio",
    display_name="MyFirstPortfolio",
    provider_name="SCAdmin",
    description="Portfolio for a project",
    message_language=servicecatalog.MessageLanguage.EN
)

Read more at Creating and Managing Portfolios.

To reference an existing portfolio into your CDK application, use the Portfolio.fromPortfolioArn() factory method:

portfolio = servicecatalog.Portfolio.from_portfolio_arn(self, "ReferencedPortfolio", "arn:aws:catalog:region:account-id:portfolio/port-abcdefghi")

Granting access to a portfolio

You can grant access to and manage the IAM users, groups, or roles that have access to the products within a portfolio. Entities with granted access will be able to utilize the portfolios resources and products via the console or AWS CLI. Once resources are deployed end users will be able to access them via the console or service catalog CLI.

import aws_cdk.aws_iam as iam

# portfolio: servicecatalog.Portfolio


user = iam.User(self, "User")
portfolio.give_access_to_user(user)

role = iam.Role(self, "Role",
    assumed_by=iam.AccountRootPrincipal()
)
portfolio.give_access_to_role(role)

group = iam.Group(self, "Group")
portfolio.give_access_to_group(group)

Sharing a portfolio with another AWS account

You can use account-to-account sharing to distribute a reference to your portfolio to other AWS accounts by passing the recipient account number. After the share is initiated, the recipient account can accept the share via CLI or console by importing the portfolio ID. Changes made to the shared portfolio will automatically propagate to recipients.

# portfolio: servicecatalog.Portfolio

portfolio.share_with_account("012345678901")

Product

Products are version friendly infrastructure-as-code templates that admins create and add to portfolios for end users to provision and create AWS resources. Service Catalog supports products from AWS Marketplace or ones defined by a CloudFormation template. The CDK currently only supports adding products of type CloudFormation. Using the CDK, a new Product can be created with the CloudFormationProduct construct. You can use CloudFormationTemplate.fromUrl to create a Product from a CloudFormation template directly from a URL that points to the template in S3, GitHub, or CodeCommit:

product = servicecatalog.CloudFormationProduct(self, "MyFirstProduct",
    product_name="My Product",
    owner="Product Owner",
    product_versions=[servicecatalog.CloudFormationProductVersion(
        product_version_name="v1",
        cloud_formation_template=servicecatalog.CloudFormationTemplate.from_url("https://raw.githubusercontent.com/awslabs/aws-cloudformation-templates/master/aws/services/ServiceCatalog/Product.yaml")
    )
    ]
)

Creating a product from a local asset

A CloudFormationProduct can also be created by using a CloudFormation template held locally on disk using Assets. Assets are files that are uploaded to an S3 Bucket before deployment. CloudFormationTemplate.fromAsset can be utilized to create a Product by passing the path to a local template file on your disk:

import path as path


product = servicecatalog.CloudFormationProduct(self, "Product",
    product_name="My Product",
    owner="Product Owner",
    product_versions=[servicecatalog.CloudFormationProductVersion(
        product_version_name="v1",
        cloud_formation_template=servicecatalog.CloudFormationTemplate.from_url("https://raw.githubusercontent.com/awslabs/aws-cloudformation-templates/master/aws/services/ServiceCatalog/Product.yaml")
    ), servicecatalog.CloudFormationProductVersion(
        product_version_name="v2",
        cloud_formation_template=servicecatalog.CloudFormationTemplate.from_asset(path.join(__dirname, "development-environment.template.json"))
    )
    ]
)

Creating a product from a stack

You can create a Service Catalog CloudFormationProduct entirely defined with CDK code using a service catalog ProductStack. A separate child stack for your product is created and you can add resources like you would for any other CDK stack, such as an S3 Bucket, IAM roles, and EC2 instances. This stack is passed in as a product version to your product. This will not create a separate CloudFormation stack during deployment.

import aws_cdk.aws_s3 as s3
import aws_cdk.core as cdk


class S3BucketProduct(servicecatalog.ProductStack):
    def __init__(self, scope, id):
        super().__init__(scope, id)

        s3.Bucket(self, "BucketProduct")

product = servicecatalog.CloudFormationProduct(self, "Product",
    product_name="My Product",
    owner="Product Owner",
    product_versions=[servicecatalog.CloudFormationProductVersion(
        product_version_name="v1",
        cloud_formation_template=servicecatalog.CloudFormationTemplate.from_product_stack(S3BucketProduct(self, "S3BucketProduct"))
    )
    ]
)

Creating a Product from a stack with a history of previous versions

The default behavior of Service Catalog is to overwrite each product version upon deployment. This applies to Product Stacks as well, where only the latest changes to your Product Stack will be deployed. To keep a history of the revisions of a ProductStack available in Service Catalog, you would need to define a ProductStack for each historical copy.

You can instead create a ProductStackHistory to maintain snapshots of all previous versions. The ProductStackHistory can be created by passing the base productStack, a currentVersionName for your current version and a locked boolean. The locked boolean which when set to true will prevent your currentVersionName from being overwritten when there is an existing snapshot for that version.

import aws_cdk.aws_s3 as s3
import aws_cdk.core as cdk


class S3BucketProduct(servicecatalog.ProductStack):
    def __init__(self, scope, id):
        super().__init__(scope, id)

        s3.Bucket(self, "BucketProduct")

product_stack_history = servicecatalog.ProductStackHistory(self, "ProductStackHistory",
    product_stack=S3BucketProduct(self, "S3BucketProduct"),
    current_version_name="v1",
    current_version_locked=True
)

We can deploy the current version v1 by using productStackHistory.currentVersion()

import aws_cdk.aws_s3 as s3
import aws_cdk.core as cdk


class S3BucketProduct(servicecatalog.ProductStack):
    def __init__(self, scope, id):
        super().__init__(scope, id)

        s3.Bucket(self, "BucketProductV2")

product_stack_history = servicecatalog.ProductStackHistory(self, "ProductStackHistory",
    product_stack=S3BucketProduct(self, "S3BucketProduct"),
    current_version_name="v2",
    current_version_locked=True
)

product = servicecatalog.CloudFormationProduct(self, "MyFirstProduct",
    product_name="My Product",
    owner="Product Owner",
    product_versions=[
        product_stack_history.current_version()
    ]
)

Using ProductStackHistory all deployed templates for the ProductStack will be written to disk, so that they will still be available in the future as the definition of the ProductStack subclass changes over time. It is very important that you commit these old versions to source control as these versions determine whether a version has already been deployed and can also be deployed themselves.

After using ProductStackHistory to deploy version v1 of your ProductStack, we make changes to the ProductStack and update the currentVersionName to v2. We still want our v1 version to still be deployed, so we reference it by calling productStackHistory.versionFromSnapshot('v1').

import aws_cdk.aws_s3 as s3
import aws_cdk.core as cdk


class S3BucketProduct(servicecatalog.ProductStack):
    def __init__(self, scope, id):
        super().__init__(scope, id)

        s3.Bucket(self, "BucketProductV2")

product_stack_history = servicecatalog.ProductStackHistory(self, "ProductStackHistory",
    product_stack=S3BucketProduct(self, "S3BucketProduct"),
    current_version_name="v2",
    current_version_locked=True
)

product = servicecatalog.CloudFormationProduct(self, "MyFirstProduct",
    product_name="My Product",
    owner="Product Owner",
    product_versions=[
        product_stack_history.current_version(),
        product_stack_history.version_from_snapshot("v1")
    ]
)

Adding a product to a portfolio

You add products to a portfolio to organize and distribute your catalog at scale. Adding a product to a portfolio creates an association, and the product will become visible within the portfolio side in both the Service Catalog console and AWS CLI. You can add a product to multiple portfolios depending on your organizational structure and how you would like to group access to products.

# portfolio: servicecatalog.Portfolio
# product: servicecatalog.CloudFormationProduct


portfolio.add_product(product)

Tag Options

TagOptions allow administrators to easily manage tags on provisioned products by providing a template for a selection of tags that end users choose from. TagOptions are created by specifying a tag key with a set of allowed values and can be associated with both portfolios and products. When launching a product, both the TagOptions associated with the product and the containing portfolio are made available.

At the moment, TagOptions can only be deactivated in the console.

# portfolio: servicecatalog.Portfolio
# product: servicecatalog.CloudFormationProduct


tag_options_for_portfolio = servicecatalog.TagOptions(self, "OrgTagOptions",
    allowed_values_for_tags={
        "Group": ["finance", "engineering", "marketing", "research"],
        "CostCenter": ["01", "02", "03"]
    }
)
portfolio.associate_tag_options(tag_options_for_portfolio)

tag_options_for_product = servicecatalog.TagOptions(self, "ProductTagOptions",
    allowed_values_for_tags={
        "Environment": ["dev", "alpha", "prod"]
    }
)
product.associate_tag_options(tag_options_for_product)

Constraints

Constraints are governance gestures that you place on product-portfolio associations that allow you to manage minimal launch permissions, notifications, and other optional actions that end users can perform on products. Using the CDK, if you do not explicitly associate a product to a portfolio and add a constraint, it will automatically add an association for you.

There are rules around how constraints are applied to portfolio-product associations. For example, you can only have a single "launch role" constraint applied to a portfolio-product association. If a misconfigured constraint is added, synth will fail with an error message.

Read more at Service Catalog Constraints.

Tag update constraint

Tag update constraints allow or disallow end users to update tags on resources associated with an AWS Service Catalog product upon provisioning. By default, if a Tag Update constraint is not configured, tag updating is not permitted. If tag updating is allowed, then new tags associated with the product or portfolio will be applied to provisioned resources during a provisioned product update.

# portfolio: servicecatalog.Portfolio
# product: servicecatalog.CloudFormationProduct


portfolio.add_product(product)
portfolio.constrain_tag_updates(product)

If you want to disable this feature later on, you can update it by setting the "allow" parameter to false:

# portfolio: servicecatalog.Portfolio
# product: servicecatalog.CloudFormationProduct


# to disable tag updates:
portfolio.constrain_tag_updates(product,
    allow=False
)

Notify on stack events

Allows users to subscribe an AWS SNS topic to a provisioned product's CloudFormation stack events. When an end user provisions a product it creates a CloudFormation stack that notifies the subscribed topic on creation, edit, and delete events. An individual SNS topic may only have a single subscription to any given portfolio-product association.

import aws_cdk.aws_sns as sns

# portfolio: servicecatalog.Portfolio
# product: servicecatalog.CloudFormationProduct


topic1 = sns.Topic(self, "Topic1")
portfolio.notify_on_stack_events(product, topic1)

topic2 = sns.Topic(self, "Topic2")
portfolio.notify_on_stack_events(product, topic2,
    description="description for topic2"
)

CloudFormation template parameters constraint

CloudFormation template parameter constraints allow you to configure the provisioning parameters that are available to end users when they launch a product. Template constraint rules consist of one or more assertions that define the default and/or allowable values for a product’s provisioning parameters. You can configure multiple parameter constraints to govern the different provisioning parameters within your products. For example, a rule might define the EC2 instance types that users can choose from when launching a product that includes one or more EC2 instances. Parameter rules have an optional condition field that allow for rule application to consider conditional evaluations. If a condition is specified, all assertions will be applied if the condition evaluates to true. For information on rule-specific intrinsic functions to define rule conditions and assertions, see AWS Rule Functions.

import aws_cdk.core as cdk

# portfolio: servicecatalog.Portfolio
# product: servicecatalog.CloudFormationProduct


portfolio.constrain_cloud_formation_parameters(product,
    rule=servicecatalog.TemplateRule(
        rule_name="testInstanceType",
        condition=cdk.Fn.condition_equals(cdk.Fn.ref("Environment"), "test"),
        assertions=[servicecatalog.TemplateRuleAssertion(
            assert=cdk.Fn.condition_contains(["t2.micro", "t2.small"], cdk.Fn.ref("InstanceType")),
            description="For test environment, the instance type should be small"
        )]
    )
)

Set launch role

Allows you to configure a specific IAM role that Service Catalog assumes on behalf of the end user when launching a product. By setting a launch role constraint, you can maintain least permissions for an end user when launching a product. For example, a launch role can grant permissions for specific resource creation like an S3 bucket that the user. The launch role must be assumed by the Service Catalog principal. You can only have one launch role set for a portfolio-product association, and you cannot set a launch role on a product that already has a StackSets deployment configured.

import aws_cdk.aws_iam as iam

# portfolio: servicecatalog.Portfolio
# product: servicecatalog.CloudFormationProduct


launch_role = iam.Role(self, "LaunchRole",
    assumed_by=iam.ServicePrincipal("servicecatalog.amazonaws.com")
)

portfolio.set_launch_role(product, launch_role)

You can also set the launch role using just the name of a role which is locally deployed in end user accounts. This is useful for when roles and users are separately managed outside of the CDK. The given role must exist in both the account that creates the launch role constraint, as well as in any end user accounts that wish to provision a product with the launch role.

You can do this by passing in the role with an explicitly set name:

import aws_cdk.aws_iam as iam

# portfolio: servicecatalog.Portfolio
# product: servicecatalog.CloudFormationProduct


launch_role = iam.Role(self, "LaunchRole",
    role_name="MyRole",
    assumed_by=iam.ServicePrincipal("servicecatalog.amazonaws.com")
)

portfolio.set_local_launch_role(product, launch_role)

Or you can simply pass in a role name and CDK will create a role with that name that trusts service catalog in the account:

import aws_cdk.aws_iam as iam

# portfolio: servicecatalog.Portfolio
# product: servicecatalog.CloudFormationProduct


role_name = "MyRole"
launch_role = portfolio.set_local_launch_role_name(product, role_name)

See Launch Constraint documentation to understand the permissions that launch roles need.

Deploy with StackSets

A StackSets deployment constraint allows you to configure product deployment options using AWS CloudFormation StackSets. You can specify one or more accounts and regions into which stack instances will launch when the product is provisioned. There is an additional field allowStackSetInstanceOperations that sets ability for end users to create, edit, or delete the stacks created by the StackSet. By default, this field is set to false. When launching a StackSets product, end users can select from the list of accounts and regions configured in the constraint to determine where the Stack Instances will deploy and the order of deployment. You can only define one StackSets deployment configuration per portfolio-product association, and you cannot both set a launch role and StackSets deployment configuration for an assocation.

import aws_cdk.aws_iam as iam

# portfolio: servicecatalog.Portfolio
# product: servicecatalog.CloudFormationProduct


admin_role = iam.Role(self, "AdminRole",
    assumed_by=iam.AccountRootPrincipal()
)

portfolio.deploy_with_stack_sets(product,
    accounts=["012345678901", "012345678902", "012345678903"],
    regions=["us-west-1", "us-east-1", "us-west-2", "us-east-1"],
    admin_role=admin_role,
    execution_role_name="SCStackSetExecutionRole",  # Name of role deployed in end users accounts.
    allow_stack_set_instance_operations=True
)