AWS Global Accelerator: A Comprehensive Cheat Sheet

What is AWS Global Accelerator?

AWS Global Accelerator is a networking service that improves the performance, availability, and security of your public-facing applications for both local and global users. It leverages the vast, congestion-free AWS global network to direct user traffic over the most optimal path to your application endpoints. By doing so, it reduces latency and jitter, and improves throughput compared to traffic routed over the public internet.

At its core, Global Accelerator provides you with two static anycast IP addresses that act as a fixed entry point to your application. These IPs are announced from multiple AWS edge locations, ensuring that user traffic gets onto the AWS network as quickly as possible.

How Global Accelerator Works: The Core Components

Global Accelerator's architecture is composed of several key components that work together to route traffic efficiently.

1. Accelerator: This is the resource you create to direct traffic to your endpoints. It includes the static IP addresses that serve as the front door to your application.

2. Listener: A listener processes inbound connections to the Accelerator based on the port (or port range) and protocol (TCP or UDP) that you configure. Each listener directs traffic to one or more endpoint groups.

3. Endpoint Group: You configure one endpoint group for each AWS Region where you have application endpoints. This group is responsible for directing traffic to the endpoints within that specific region. You can control the percentage of traffic that a region receives using a "traffic dial".

4. Endpoint: These are the specific AWS resources that serve your application traffic. Endpoints can be Network Load Balancers, Application Load Balancers, EC2 Instances, or Elastic IP addresses.

The flow is as follows: User Traffic -> Static IP -> Accelerator -> Listener -> Regional Endpoint Group -> Healthy Endpoint.

Global Accelerator vs. Amazon CloudFront

This is a critical comparison, as both services use the AWS global edge network but solve different problems.

| Feature | AWS Global Accelerator | Amazon CloudFront |

| :--- | :--- | :--- |

| Primary Use Case | Improves performance for any TCP/UDP application (non-HTTP and HTTP). Optimizes the network path. | Caches HTTP/S content at the edge to reduce latency and origin load. |

| Caching | No caching. It is a proxy service that routes requests to the origin. | Yes. Its primary function is to cache static and dynamic web content. |

| Protocols | TCP and UDP. | HTTP and HTTPS. |

| Entry Point | Provides two static anycast IP addresses. | Provides a unique domain name (e.g., d1234.cloudfront.net). |

| Good For | Gaming, VoIP, IoT, live video, financial applications, and any latency-sensitive TCP/UDP workload. | Websites, video on demand, API acceleration, software distribution, and other web-based content. |

Rule of Thumb: Use CloudFront for web applications and content delivery that benefits from caching. Use Global Accelerator for non-HTTP workloads or latency-sensitive applications where you need to optimize the network path and provide a static IP entry point. They can also be used together.

Key Features and Benefits

• Performance Improvement: By routing traffic over the AWS private global network instead of the public internet, Global Accelerator can significantly reduce first-byte latency and improve throughput.

• High Availability and Fault Tolerance: It automatically detects an unhealthy endpoint and instantly reroutes traffic to the next nearest healthy endpoint in another Availability Zone or Region, typically enabling failover in under a minute.

• Static Anycast IP Addresses: You get a fixed set of IP addresses that do not change. This simplifies firewall whitelisting and allows you to move or change backend resources without any impact on your users or client-facing configurations.

• Client Affinity: For stateful applications, you can configure client affinity to ensure that all requests from a specific user are consistently directed to the same endpoint to maintain session state.

• Bring Your Own IP (BYOIP): You can use your own publicly routable IP address range with Global Accelerator.

Common Use Cases

• Gaming: Provide a low-latency, stable connection for global players.

• Voice over IP (VoIP) and Video: Ensure high-quality, real-time communication by minimizing jitter and latency.

• Internet of Things (IoT): Improve the reliability and speed of device-to-cloud communication.

• API Acceleration: For HTTP/S APIs that do not benefit from caching but require the highest possible availability and performance.

Security and Compliance

• DDoS Protection: Integrates with AWS Shield Standard at no additional cost, providing always-on protection against common infrastructure-level DDoS attacks.

• Compliance: Meets major compliance standards, including SOC, PCI, HIPAA, GDPR, and ISO.

• VPC Security: By routing traffic through the AWS network, it helps shield your application endpoints in your VPC from direct internet exposure.

Pricing Model

Global Accelerator has two cost components:

1. Fixed Hourly Fee: A small, fixed fee for each hour that an accelerator is provisioned in your account.

2. Data Transfer-Premium (DT-P): An incremental charge for data transferred over the AWS network, billed per gigabyte. This fee is on top of standard EC2 data transfer rates.