Home Arrow Icon Knowledge base Arrow Icon Global Arrow Icon What are the benefits of using a custom routing accelerator over a standard accelerator


What are the benefits of using a custom routing accelerator over a standard accelerator


Using a custom routing accelerator in AWS Global Accelerator offers several distinct benefits over a standard accelerator, particularly for applications requiring deterministic, fine-grained control over routing user traffic to specific Amazon EC2 instances.

Benefits of Custom Routing Accelerators

**1. Deterministic Traffic Routing to Specific EC2 Instances and Ports
Unlike standard accelerators, which automatically route traffic to the nearest healthy endpoint based on geographic proximity and endpoint health, custom routing accelerators allow you to use your own application logic to route traffic to a specific EC2 instance and port. This deterministic routing is essential for use cases where multiple users must be assigned to a particular server or session, such as multiplayer gaming, VoIP, EdTech, or social media applications that require users to connect to a specific media server or game session based on factors like geographic location, player skill, or session configuration[2][3].

**2. Integration with Custom Application Logic
Custom routing accelerators enable your application to retrieve a static mapping of accelerator ports to destination EC2 private IP addresses and ports. This mapping allows your matchmaking or session management services to precisely route users to the correct backend instance without modifying client software. This integration facilitates complex routing decisions based on your application’s unique requirements, which standard accelerators cannot support[2][3].

**3. Improved Security and Simplified Network Architecture
With custom routing accelerators, you avoid exposing individual EC2 instances directly to the internet, which can pose security risks. Instead of assigning Elastic IPs to each instance, the accelerator provides static anycast IP addresses that route traffic internally within AWS’s global network to the appropriate subnet and instance port. This reduces the attack surface and simplifies IP management[2].

**4. Leveraging AWS Global Network Performance
Custom routing accelerators still leverage the AWS global network infrastructure to route traffic, ensuring low latency, jitter, and packet loss. This means you gain the performance and availability benefits of Global Accelerator while maintaining precise control over traffic routing[2][3].

**5. Support for Multiple Protocols and Flexible Port Mapping
Custom routing accelerators allow you to specify TCP, UDP, or both protocols at the endpoint group level, providing flexibility for applications that require different transport protocols for different types of traffic. You can define port ranges that map directly to your EC2 instances, enabling consistent port usage across multiple instances[2][3].

**6. Scalability for Complex Use Cases
Custom routing accelerators can handle routing from a few to thousands of users to specific EC2 endpoints, making them suitable for large-scale applications that require session-based user assignment, such as large multiplayer games or real-time communication platforms[2].

Summary

While standard AWS Global Accelerators provide automatic, health- and proximity-based routing suitable for general load balancing and failover scenarios, custom routing accelerators offer enhanced control by enabling deterministic routing to specific EC2 instances and ports. This capability is crucial for applications needing session affinity, custom matchmaking, or specialized routing logic, all while benefiting from AWS’s global network performance and security features.

This makes custom routing accelerators ideal for:

- Multiplayer gaming with session-based player assignment
- VoIP and real-time communication requiring specific media server connections
- Applications needing fine-grained control over traffic routing beyond geographic proximity or endpoint health
- Scenarios where exposing individual EC2 instances directly to the internet is undesirable

In essence, custom routing accelerators combine the performance and availability benefits of AWS Global Accelerator with the flexibility to implement complex, application-specific routing logic[2][3][4].

Citations:
[1] https://aws.amazon.com/blogs/networking-and-content-delivery/introducing-aws-global-accelerator-custom-routing-accelerators/
[2] https://docs.aws.amazon.com/global-accelerator/latest/dg/about-custom-routing-how-it-works.html
[3] https://cloudchipr.com/blog/aws-global-accelerator
[4] https://www.nops.io/glossary/what-is-aws-global-accelerator/
[5] https://repost.aws/questions/QUzbDXdC12QDy-igAdhLPi7g/standard-global-accelerator-ec2-instance-endpoint-which-eni-does-the-traffic-go-to
[6] https://www.youtube.com/watch?v=DuXx0UpUKSg
[7] https://docs.aws.amazon.com/global-accelerator/latest/dg/about-custom-routing-accelerators.html
[8] https://www.reddit.com/r/aws/comments/z1vazp/global_accelerator_routing_to_another_accounts_alb/
[9] https://boto3.amazonaws.com/v1/documentation/api/1.17.58/reference/services/globalaccelerator.html