View a markdown version of this page

Multi-Region and hybrid architecture for low-latency games - Games Industry Lens

Multi-Region and hybrid architecture for low-latency games

This section describes a multi-Region and hybrid architecture for low-latency games.

Reducing latency with network acceleration and game servers deployed globally

Reducing latency with network acceleration and game servers deployed globally

  1. Players in a globally available game can originate from anywhere. When a player requests a game session or match, their game client sends a request to the game backend service registered with Amazon Route 53. Route 53 latency-based routing can be used to route the player to the closest available game backend.

  2. The game backend is deployed in multiple AWS Regions closest to player populations. Each game backend includes a Regional matchmaking service that finds a game session from across the game Regions. Although a player's matchmaking request is processed by a Regional matchmaking service near them, the matchmaking service is capable of routing a player to a game session in another game Region, if necessary. This action improves resiliency and performance. Additionally, each game backend service uses AWS WAF and AWS Shield Advanced to provide layer-7 web filtering and bot control, and protections against distribution denial of service (DDoS) attacks. You have many options for how you build your game backend service, such as serverless, containers, EC2 instances, or hosting the game backend service in your own data centers.

  3. To improve the experience for players by reducing network latency and jitter, a custom routing accelerator is deployed using AWS Global Accelerator, which automatically optimizes the routing of traffic from the game client to the game server using the AWS global network. You configure the custom routing accelerator to map Global Accelerator listener ports to your game server's EC2 instance port. The game client connects to the Global Accelerator IP and port that serves as a proxy and deterministically routes the players to the correct game server IP and port that is hosting the game session.

  4. Your game includes player-friendly logical game Regions that represent a collection of game server hosting locations that are geographically close to each other, such as North America or Asia Pacific. To reduce latency and increase geographic coverage, a combination of different game server hosting solutions can be used to improve the player experience. Prioritize the use of AWS Regions where possible, as these locations are fully featured and contain the largest capacity footprint.

  5. Use AWS Local Zones to host game servers in geographic locations of under-served players where you do not have existing hosting facilities, or an AWS Region is not available.

  6. Deploy AWS Outposts into your existing on-premises data centers and co-location providers to create a seamless control plane and management experience across each deployment location using fully managed racks and rack-mounted servers. AWS Outposts are also beneficial if you don't have existing server capacity in your on-premises environment. However, if you want a hybrid implementation with software running on your own existing server infrastructure, you should use Amazon EKS Anywhere, which allows you to create and run Kubernetes clusters on your own infrastructure with connectivity back to Amazon EKS. This provides a consistent console view of your Kubernetes clusters in on-premises.