Skip to main content

Amazon GameLift Streams

Page topics

General

Open all
Amazon GameLift Streams helps game developers and publishers deliver game streaming experiences at up to 1080p resolution and 60 frames per second (fps) across multiple devices including TVs, PCs, mobile phones, and tablets. Using a single AWS capability, publishers can deploy their game content in minutes, without modifications, onto fully-managed cloud-based GPU instances and deliver them through the AWS Network Backbone directly to any end-player device with a web browser. Amazon GameLift Streams enables users to start playing games in just a few seconds, without waiting for a download or an install, delivering a gameplay experience that is nearly indistinguishable from playing the game locally on a PC or gaming console. With Amazon GameLift Streams, game publishers can distribute their existing and new games direct-to-players, without managing infrastructure and paying only for what they allocate. Publishers also maintain full control over the player relationship, player experience, branding, and business model.
Game streaming through Amazon GameLift Streams offers several key advantages. First, it enables you to deliver smooth, real-time, high frame rate gameplay globally to nearly any device, and without the need for gamers to have a local GPU. With Amazon GameLift Streams, you can stream your games in minutes, without the need to modify your game code to make it compatible with the latest devices and browser versions. It helps you reach more players directly with your own distribution channels, and you and your teams can save hours of playtesting setup time since game builds are not installed locally.
Amazon GameLift Streams uses the Web Real-Time Communication (WebRTC) open standard protocol, which helps you stream your game without modifications. Amazon GameLift Streams technology is specifically designed to deliver real-time gameplay experiences with minimal player-to-cloud latency. It operates from multiple AWS Regions globally, allowing you to deploy compute resources closer to your players to minimize latency. While some minimal latency is inherent in any streaming solution, Amazon GameLift Streams is optimized for responsive gameplay.
Amazon GameLift Streams supports direct streaming of Windows and Linux-based games with little to no modification required. It is compatible with major game engines, including Unreal, Unity, Godot, and custom game engines.
Amazon GameLift Streams is available in the following AWS Regions: US East (N. Virginia), US East (Ohio), US West (Oregon), Asia Pacific (Tokyo), Europe (Frankfurt), and Europe (Ireland).

Amazon GameLift Streams offers two types of scaling through stream capacity — the maximum number of concurrent streams that can be active at a time:

  • Always-on capacity: Pre-allocated resources that are ready to handle stream requests without delay.
  • On-demand capacity: Resources that are allocated in response to stream requests and de-allocated when sessions terminate.

You can adjust both types of capacity at any time to meet changes in user demand. If you provision a mix of the two in a single stream group, Amazon GameLift Streams fulfills streaming requests using always-on capacity first, then provisions on-demand capacity as needed.
Players only need a compatible web browser and a stable internet connection with at least 10 Mbps bandwidth for 1080p streaming. Amazon GameLift Streams supports major browsers, including Google Chrome, Microsoft Edge, Firefox, and Safari. For optimal experience, we recommend using Chrome, Edge, or a custom Chromium-based desktop application.
Yes, Amazon GameLift Streams provides demo materials in the GameLift Streams Web SDK bundle, available through the Getting Started page. This includes an example web client and back-end streaming capability that you can use to test the capability and understand how to implement streaming features.
No, you are not required to authenticate players. However, you must provide a unique, non-personally identifiable player ID for each player. While players don't need to authenticate, anything to do with uploading games and requesting stream sessions should be authenticated from your backend system using your AWS credentials. For information about setting up AWS credentials, refer to the "Setting up Amazon GameLift Streams" section in the developer guide.
Troubleshooting features offered by Amazon GameLift Streams include detailed logging, performance metrics through Amazon CloudWatch, and API call tracking through AWS CloudTrail. It provides visibility into stream group capacity, usage, performance, resource utilization, and stream status. Additionally, you can export stream session files for debugging purposes.
Amazon GameLift Streams provides near real-time metrics through Amazon CloudWatch. However, If you are using the sample web application, you can access WebRTC metrics as they happen in real time. Metrics for stream group capacity and usage are published every minute, allowing you to monitor performance, resource utilization, and user engagement in near real-time. This enables quick identification and resolution of potential issues while maintaining optimal service delivery.
Yes, Amazon GameLift Streams integrates with AWS monitoring services to provide operational oversight. You can use Amazon CloudWatch to monitor metrics, create customized dashboards, and set alarms that notify you when specified thresholds are reached. Additionally, you can use AWS CloudTrail to track API activity and AWS Billing and Cost Management to monitor costs and usage. These tools help you maintain operational health and respond quickly to any issues.
During development, Amazon GameLift Streams offers several features to support rapid iteration. You can test streams directly in the Amazon GameLift Streams console. It also provides a sample back-end capability and web client in the Amazon GameLift Streams Web SDK bundle for quick implementation testing. You can use on-demand capacity or scale stream groups to zero capacity when not in use to manage costs during development. To get started and iterate quickly, select Windows for ultimate game compatibility; then test with Proton to achieve multi-tenancy at scale.
For production environments, Amazon GameLift Streams enables you to update applications and stream groups independently. You can create new applications while maintaining existing ones, test new configurations in separate stream groups, and gradually migrate traffic to new versions. It also supports multi-application stream groups, enabling you to manage multiple versions of your game through a single set of compute resources.
An Amazon GameLift Streams application is a resource that combines a read-only copy of your game files, as originally imported from Amazon Simple Storage Service (S3), with metadata that specifies the runtime environment and the executable that you want to run when streaming your game.
A runtime environment in Amazon GameLift Streams is the underlying operating system and software environment that runs your game on the Amazon GameLift Streams capability. Available options include Microsoft Windows Server 2022 Base, Ubuntu 22.04 LTS, and Proton (versions 8.0-5 and 8.0-2c). Each runtime environment offers different compatibility and performance characteristics, allowing you to choose the best option for your game. To get started and iterate quickly, select Windows for ultimate game compatibility; then test with Proton to achieve multi-tenancy at scale.
A stream class in Amazon GameLift Streams defines the compute resources used to run and stream your game. Each stream class offers different combinations of CPU, GPU, RAM, and other specifications to meet your performance needs and budget requirements. Amazon GameLift Streams offers NVIDIA GPU-based stream classes with "high" (shared) and "ultra" (dedicated) options. The stream class you choose directly impacts both streaming quality and cost.

Stream classes primarily differ in their resource allocation and tenancy:

  • "Ultra" classes (like gen5n_ultra) provide dedicated resources with higher specifications (8 vCPUs, more VRAM and RAM) and 1 GPU per stream session. Start here for the most graphically intensive games that need the most processing power.
  • "High" classes (like gen5n_high) offer shared resources with moderate specifications (4 vCPUs) and the GPU is shared between up to two concurrent stream sessions. Start here for games that require only moderate graphical power that have the potential to host more than one game per stream class.
  • Windows-specific classes (like gen5n_win2022) are specially configured for Windows games with DirectX support.

Stream groups

Open all
A stream group represents a collection of compute resources that Amazon GameLift Streams uses to stream your applications to end users. When you create a stream group, you specify the stream class, default application, stream capacity, and geographical locations where your game can run. Stream groups manage how Amazon GameLift Streams allocates resources and handles concurrent streams, allowing you to effectively manage capacity and costs.
Stream capacity represents the maximum number of concurrent streams that can be active at a time.
A multi-location stream group can host games and stream sessions from multiple geographical locations, in addition to the primary location (the AWS Region where you created the stream group). This allows you to stream content from locations closer to your end users for lower latency while managing capacity from a single pool. You can manage capacity independently for each location within the stream group.
A multi-application stream group has links to multiple applications, enabling you to stream different applications using the same pool of compute resources. This many-to-one relationship allows you to efficiently manage resources across multiple applications or versions of your game. When starting a stream session, you specify which linked application to stream, supporting faster iterations and shared capacity between multiple game versions or different content offerings.
When a stream session ends, Amazon GameLift Streams automatically removes any new files generated during the session. Refer to the shared security model documentation. You can optionally configure the capability to export stream session files to an Amazon Simple Storage Service (S3) bucket before termination, which is useful for debugging or verification purposes.
While you can update certain aspects of a live stream group (such as capacity and general settings), some configurations cannot be changed after creation, such as the stream class. As a best practice, rather than modifying a production stream group significantly, we recommend creating a new stream group with the desired changes and gradually migrating traffic to it. This approach provides a safer rollback path if issues arise.

Amazon GameLift Streams provisions capacity at the locations you have selected. The locations indicate where GameLift Streams hosts your application and streams sessions. By default, you can stream from the same location where you created your stream group, known as the primary location. Additionally, a stream group can extend its coverage to stream from other supported locations. You can choose from the following AWS Regions to be your primary location: US East (Ohio), US West (Oregon), Asia Pacific (Tokyo), and Europe (Frankfurt).

Each primary location can also extend coverage to additional locations for lower latency streaming. In addition to the AWS Regions listed above, we offer two remote locations that can be managed from your primary location: US East (N. Virginia) and Europe (Ireland).
No, you cannot change the stream class after creating a stream group. If you need to use a different stream class, you must create a new stream group with the desired stream class and migrate your streaming traffic to the new group.
Yes, Amazon GameLift Streams exclusively uses accelerated GPU stream classes. It offers NVIDIA-based options using NVIDIA A10G Tensor or T4 Tensor GPUs options, allowing you to choose the best hardware configuration for your game's graphical requirements.
Always-on and on-demand stream capacity serve different needs in your streaming infrastructure. Always-on capacity provides immediate availability for stream requests with no startup delay, making it ideal for scenarios requiring quick response times and predictable workloads. On- demand capacity offers cost optimization by provisioning resources only when needed, typically taking under 5 minutes to start up. This makes it particularly suitable for variable workloads where immediate availability isn't critical. You can enable both always-on and on-demand in the same stream group.

Compatibility

Open all
Amazon GameLift Streams is optimized to support a wide range of game genres that can accommodate streaming latency. It delivers high-quality streaming up to 1080p resolution and 60 frames per second, making it particularly effective for strategy games, role-playing games, adventure games, and simulation games. While Amazon GameLift Streams can support fast-paced genres, we recommend thorough testing to ensure the streaming performance aligns with your specific gameplay requirements.
Amazon GameLift Streams supports multiple server environments to accommodate different development needs. It supports Microsoft Windows Server 2022 Base, Ubuntu 22.04 LTS, and Proton. Proton is a game-optimized compatibility layer that enables Windows software to run on a Linux- based runtime environment. You can develop applications in any programming language compatible with these environments, and it provides specific optimizations for game engines such as Unreal Engine.
Amazon GameLift Streams provides a Web SDK bundle, which is the recommended starting point for developing your own GameLift Streams solution. This bundle contains the JavaScript AWS SDK with GameLift Streams APIs, a sample web server, and a sample web client for connecting to streams. For developers who prefer other programming languages, Amazon GameLift Streams offers custom versions of the AWS SDK that include Amazon GameLift Streams APIs in JavaScript <PHP, C#, Go, and C++?>. All streaming content is delivered through web browsers, making it accessible across devices that support modern web browsers.
Amazon GameLift Streams provides gamepad support through the Web Gamepad API, with optimal compatibility when using Google Chrome, Microsoft Edge, or custom Chromium-based applications. It's important to note that at this time, gamepad support isn't available when using the Ubuntu 22.04 LTS runtime environment.
Amazon GameLift Streams supports microphone input on Windows and Proton runtime environments through Chrome, Edge, and Firefox. On Mac, microphone input is supported through Chrome, Edge, Safari, and Firefox. On mobile devices, microphone input is supported through their respective supported browsers. This allows developers to implement voice-based features in their streaming applications. It's important to note that at this time, microphone support isn't available when using the Ubuntu 22.04 LTS runtime environment.
Amazon GameLift Streams uses Microsoft Windows Server 2022 Base as its Windows runtime environment. This platform provides support for DirectX 12 and DirectX 11, enables compatibility with Unreal Engine 5.x applications, and supports both 32-bit and 64-bit applications as well as anti-cheat technology. The environment is specifically optimized for use with gen5n_win2022 and gen4n_win2022 stream classes.
Amazon GameLift Streams employs Ubuntu 22.04 LTS as its Linux runtime environment. This distribution works seamlessly with NVIDIA-based stream classes (gen5n and gen4n series), providing a robust platform for Linux-based game streaming. The environment supports both high (shared) and ultra (dedicated) configurations, although it's important to note that microphone and gamepad functionality is not available on this runtime.

Streaming

Open all
Amazon GameLift Streams uses WebRTC as its streaming protocol. This technology enables real-time, low-latency communication while maintaining secure, encrypted connections. The protocol allows direct browser-based streaming without requiring additional plugins, and supports a variety of input methods including keyboard, mouse, and gamepad interactions.
For optimal streaming performance, Amazon GameLift Streams requires a stable internet connection with at least 10 Mbps bandwidth for 1080p streaming. To minimize latency, it allows you to deploy compute resources in multiple locations closest to your players. We recommend choosing stream group locations that are geographically near your players and typically provide 100 ms or less latency to the majority of them, to ensure the best possible streaming experience.

Service quotas

Open all
By default, Amazon GameLift Streams allows you to maintain up to two stream groups per AWS account, per AWS Region. This quota is adjustable based on your needs. You can reach out to your AWS Account Manager or contact AWS support to discuss increasing this limit. This limitation helps ensure optimal resource allocation and service performance across all customers.
Yes, Amazon GameLift Streams implements a limit of 100 links in multi-application stream groups. This means a single stream group can link to up to 100 applications, and conversely, a single application can be linked to up to 100 stream groups. This quota is adjustable; reach out to your AWS Account Manager or contact AWS support.
Yes, Amazon GameLift Streams implements default stream capacity limits per AWS account, per location, which vary by stream class. To check your current quota for a specific location and stream class, refer to Service Quotas in the AWS console. These quotas can be adjusted by reaching out to your AWS Account Manager or contacting AWS support.

Billing

Open all
You pay for the compute resources and storage that you provision for your game streams, without monthly or annual contracts. Please see the Amazon GameLift Streams pricing page for more information.
Billing for Amazon GameLift Streams begins when stream capacity is allocated in a stream group (meaning compute resources are ready to stream your application) and continues as long as that stream group’s allocated capacity is maintained. You are charged for stream capacity per second, regardless of whether that allocated capacity is actively streaming or idle. Billing ends when you reduce the allocated capacity in your stream group to zero, or delete the stream group entirely.
No. At this point, we do not support the ability to bring your own licenses.

Security

Open all
Yes, Amazon GameLift Streams provides encryption protection. Application content stored in Amazon Simple Storage Service (S3) buckets is secured with server-side encryption. All API calls to GameLift Streams are made over secure SSL connections and authenticated using AWS Signature Version 4?. Additionally, it ensures secure communication between stream clients and stream servers hosted by Amazon GameLift Streams.
Amazon GameLift Streams integrates with AWS Identity and Access Management (AWS IAM) to control access to your resources. You can create IAM policies that define permissions for users, roles, and groups, specifying who can perform which actions on Amazon GameLift Streams resources. It supports identity-based policies and provides specific condition keys for fine-grained access control, allowing you to implement the principle of least privilege for your streaming infrastructure.
No.
Yes, Amazon GameLift Streams supports anti-cheat technology when using Windows-based stream classes (gen5n_win2022 and gen4n_win2022). These stream classes run on Microsoft Windows Server 2022 Base and are specifically compatible with anti-cheat software, allowing you to implement additional security measures for your games.
Stream groups require periodic re-creation to receive new capability updates, security patches, and bug fixes. As a best practice, AWS recommends re-creating stream groups every 3-4 weeks. When a stream group becomes outdated, you'll see a "Maintenance required" message in the console. Stream groups over 180 days old cannot accept new application links until they are re-created. You can use the "Create stream group" button in the maintenance message to streamline the re-creation process.

Service level agreement

Open all
See the SLA page for more information.
No.
Getting started with Amazon GameLift Streams is a straightforward process. First, complete the one-time setup tasks, which includes creating an AWS account and setting up proper credentials. Then, upload your application to an Amazon Simple Storage Service (S3) bucket, create a Amazon GameLift Streams application, configure a stream group, and test your stream directly in the console. Refer to the Getting Started page for more details.
To import your game to Amazon GameLift Streams, first upload your game files to an Amazon Simple Storage Service (S3) bucket in an uncompressed folder format. Then, create an Amazon GameLift Streams application resource by specifying the S3 URL to your application files and the executable launch path, and choosing an appropriate runtime environment. Once your application is in "Ready" status, you can safely remove the original files from your S3 bucket to avoid continual storage charges for these files.
Yes, our WebSDK and Sample Web Application are designed to integrate seamlessly into existing web pages, and there is no AWS branding on any part of the customer experience. Amazon GameLift Streams also allows you to customize both loading screens and background images. For loading screens, you can implement custom backgrounds and logos through the Amazon GameLift Streams Web SDK sample client. For background images, you can display a custom image when your application hasn't launched yet or has been exited by placing a file named "Background.bmp" in your application's Config folder. The background image should match the stream resolution of 1920 x 1080 pixels for optimal display.
To ensure a successful launch on Amazon GameLift Streams, begin preparation 6-8 weeks in advance. Notify the Amazon GameLift Streams team about your launch timeline and expected peak concurrent streams so they can ensure adequate service limits. Conduct thorough compatibility testing across multiple stream classes and runtimes, reserve capacity if needed for large-scale events, and perform load testing of your APIs and configurations. Create your final application resources and stream groups 2-3 days before launch, and maintain consistent stream groups throughout the launch event for easier troubleshooting. For more details, refer to the Launch checklist topic in the Amazon GameLift Streams Developer Guide.