AWS AppStream 2.0 with Azure Entra ID for a Dynamic Virtualized Application

Mr. Exojo

Introduction

This summer, I had the opportunity to collaborate with an IT consulting firm to address a unique challenge posed by one of their online education clients. The client sought a comprehensive proof of concept (PoC) that involved virtualizing a Windows-based application using AWS AppStream 2.0, integrated seamlessly with Azure Entra ID (formerly known as Azure Active Directory) via SAML authentication. The ultimate goal was to create a secure and personalized user experience where directory parameters are dynamically retrieved and utilized to customize the application based on incoming requests.

Embarking on this project was more complex than it might initially seem. It required in-depth research beyond the initial project scope and budget, necessitating adaptations to the architectural scheme and the addition of components and specific configurations. Moreover, the client requested the implementation of a dynamic URL configuration to extract specific parameters, ensuring that the application delivers tailored data based on each user’s session.

In this blog post, I’ll delve into the critical aspects of this PoC, highlighting the challenges faced, solutions implemented, and key takeaways from this intricate project.


1. AWS AppStream 2.0 Configuration: Image Builder, Images, Fleet, and Stack

Understanding the core components of AWS AppStream 2.0 is crucial for an effective deployment. The service primarily revolves around four components:

  • Image Builder
  • Image
  • Fleet
  • Stack

Image Builder and Image Creation

The Image Builder is a managed Amazon EC2 instance that provides a secure environment to install and configure the applications you want to stream to your users. In this project, the Image Builder was used to:

  • Install the Windows-based application.
  • Configure necessary settings and dependencies.
  • Optimize the application for streaming performance.

Time Considerations

One of the challenges encountered was the time required for each Image Builder creation and image snapshot process. Each cycle could take upwards of 20 minutes or more, depending on the complexity of the application and configuration steps involved. This latency is common in AppStream 2.0 and necessitates precise planning to prevent unnecessary delays in the implementation timeline. To mitigate this, we:

  • Scheduled image creation during off-peak hours.
  • Automated configuration tasks using scripts.
  • Thoroughly tested configurations in a development environment before applying them in the Image Builder.

Fleet and Stack: Key Concepts

The Fleet consists of the streaming instances that deliver applications to users. Fleets can be configured with different instance types and sizes based on performance requirements. For this project:

  • We configured a fleet capable of scaling based on user demand.
  • Implemented session scripts to retrieve user attributes upon login.

The Stack ties together the fleet, user access policies, and storage configurations, serving as the interface that users interact with. Key configurations included:

  • Setting up SAML 2.0 authentication to integrate with Azure Entra ID.
  • Defining user access policies to restrict application access based on user roles.
  • Configuring application settings to accept dynamic parameters from user sessions.

2. Azure Entra ID and SAML Integration

A core requirement was integrating AWS AppStream 2.0 with Azure Entra ID using SAML 2.0 authentication. This integration allows for:

  • Single Sign-On (SSO) capabilities.
  • Retrieval of user attributes such as roles, group memberships, and custom claims.
  • Enhanced security through federated authentication.

Configuration Steps

  • Setting up an Enterprise Application in Azure Entra ID: We registered AWS AppStream 2.0 as an enterprise application to enable SAML-based SSO.
  • Configuring SAML Assertions: Customized the SAML token attributes to include necessary user information.
  • Establishing Trust Relationship: Created an IAM SAML identity provider in AWS using the metadata XML file from Azure Entra ID.
  • Mapping Roles and Policies: Defined IAM roles with policies that correspond to user roles in Azure Entra ID.

IAM Role Trust Policy

The key configuration to enable this integration involved setting up IAM roles with specific policy actions required by the claims used to retrieve particular information about an Azure Entra ID user. The trust policy looked like this:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Principal": {
                "Federated": "arn:aws:iam::<AWS_ACCOUNT_ID>:saml-provider/<PROVIDER_NAME>"
            },
            "Action": [
                "sts:AssumeRoleWithSAML",
                "sts:TagSession",
                "sts:SetSourceIdentity"
            ],
            "Condition": {
                "StringEquals": {
                    "SAML:sub_type": "persistent"
                }
            }
        }
    ]
}

This policy allows users authenticated via SAML to assume the IAM role, tag their session with specific attributes, and set the source identity, ensuring that the application can retrieve and utilize user-specific parameters securely.


3. VPC and IAM Setup for AppStream

A vital component of the AppStream deployment is the Virtual Private Cloud (VPC) configuration and IAM role management.

VPC Configuration

  • Subnets: AppStream requires at least two subnets in different Availability Zones for high availability.
  • Security Groups: Configured to allow necessary inbound and outbound traffic.
  • Network Connectivity: Ensured that the streaming instances could connect to required services and resources.

IAM Roles and Policies

  • AppStream Service Role: Granted permissions necessary for AppStream to interact with AWS services.
  • Fleet Instance Role: Assigned to the fleet instances to allow them to access AWS resources securely.
  • SAML-Enabled IAM Role: Configured with trust policies to allow federated access via Azure Entra ID.

SAML Role Trust Policies

Meticulous attention was given to the trust policies associated with IAM roles to ensure:

  • Only authenticated users from Azure Entra ID could assume the roles.
  • Proper permissions were granted based on user roles and attributes.
  • Compliance with the principle of least privilege for enhanced security.

4. Implementing Dynamic URL Configuration for User-Specific Parameters

One of the client’s unique requirements was the implementation of a dynamic URL configuration to extract specific parameters, allowing the application to deliver tailored data based on each user’s session.

Approach

  • Session Scripts: Implemented session scripts that run when a user connects to the streaming instance. These scripts retrieve user attributes and pass them to the application.
  • Custom Parameters: Configured the application to accept command-line arguments or environment variables for dynamic data input.
  • URL Redirection: Set up mechanisms to capture URL parameters from the user’s access point and relay them to the application.

Challenges

  • Ensuring secure handling of user data and parameters.
  • Maintaining the integrity of the application while accepting dynamic inputs.
  • Synchronizing the retrieval of SAML assertions with application startup.

5. Technical Challenges and Implementation Timelines

Image Creation Latency

The time required for creating image builders and images was significant. To address this:

  • Optimized the image creation process by minimizing unnecessary software installations.
  • Used scripts to automate repetitive configuration tasks.
  • Maintained a version-controlled repository of configurations to expedite future iterations.

SAML and Azure Entra ID Integration

Integrating SAML authentication required multiple iterations to ensure:

  • Accurate mapping of user roles and attributes.
  • Secure and seamless sign-in experience.
  • Proper configuration of assertions and claims in the SAML token.

Dynamic Parameter Handling

Implementing the dynamic URL configuration posed challenges such as:

  • Parsing and sanitizing user input to prevent security vulnerabilities.
  • Ensuring compatibility between the application’s expected input format and the parameters provided.

6. Conclusion: Key Takeaways and Lessons Learned

This project was a comprehensive exploration into the intricacies of AWS AppStream 2.0 and its interaction with external identity providers like Azure Entra ID. Key takeaways include:

  • Planning for Latency: Recognize and plan for inherent latencies in services like AppStream’s Image Builder to optimize timelines.
  • Meticulous IAM Configuration: Correctly configuring IAM roles and policies is critical, especially when dealing with federated authentication and sensitive user attributes.
  • Network and Security Considerations: Ensuring that the VPC and security groups are correctly set up is essential for both connectivity and security.
  • Dynamic Configuration Management: Implementing dynamic parameters requires careful handling to maintain security while providing personalized user experiences.
  • Iterative Testing and Validation: Multiple iterations and tests are often necessary to fine-tune integrations and configurations, emphasizing the need for a flexible project scope and timeline.

Final Thoughts

Through careful planning, attention to detail, and collaborative problem-solving, we successfully delivered a practical proposal that met the client’s needs within the technical constraints of the environment. This project not only enhanced the client’s capabilities but also provided valuable insights into integrating AWS and Azure services to create robust, secure, and dynamic solutions.


Additional Resources

Feel free to reach out if you have any questions or need further clarification on any of the topics discussed in this post.

Userful? thanks for share it