AWS Networking Fundamentals

Edge Networks

Global Accelerator

In this lesson, we’ll dive into AWS Global Accelerator—its purpose, architecture, and how it contrasts with Amazon CloudFront. By the end, you’ll understand how Global Accelerator leverages AWS’s private backbone to reduce latency, improve throughput, and deliver a more consistent application experience.

The Challenge: Internet Latency and Unpredictability

When your application is hosted in a North America region, a user request from anywhere in the world must:

  1. Travel from the user’s device through their ISP
  2. Traverse a variable, often congested path across the public internet
  3. Reach your application’s endpoint

Because the public internet is inherently unpredictable, you may encounter high latency, packet loss, and inconsistent performance.

Introducing AWS Global Accelerator

AWS Global Accelerator solves these issues by assigning your application two static anycast IP addresses (or you can bring your own IP addresses). These IPs are announced from AWS edge locations in major metro areas worldwide, ensuring user traffic enters the AWS global network as close to the user as possible.

How It Works

  1. AWS provisions two static anycast IP addresses for your accelerator.
  2. Each IP is announced from multiple AWS edge locations.
  3. A user’s request is routed to the nearest edge location using the anycast IP.
  4. From that edge, traffic travels over the AWS private global network backbone to your endpoint (Application Load Balancer, Network Load Balancer, EC2 instance, or Elastic IP).

The image illustrates a map showing the AWS Backbone Network with various global locations marked by icons, connected by lines. It is labeled "Global Accelerator" and is copyrighted by KodeKloud.

Key Benefits

BenefitDescription
Lower, Consistent LatencyTraffic stays on AWS’s optimized backbone, avoiding public internet hops
Improved Throughput & Packet LossPrivate network reduces congestion and retransmissions
Automatic Health Checks & FailoverIntelligent routing directs users away from unhealthy or degraded endpoints
Static Anycast IPsSimplify DNS management and eliminate the need for regional endpoint records

Note

You can bring your own IP addresses (BYOIP) to Global Accelerator for seamless migration and IP ownership continuity.

Traffic Flow Example

A client in Europe sends a request to one of your accelerator’s anycast IPs.

  1. The request is picked up by the nearest AWS edge location (e.g., Frankfurt).
  2. It rides the AWS private backbone across regions (e.g., via AWS Points of Presence).
  3. It terminates at your origin—whether that’s an Application Load Balancer in us-east-1 or a Network Load Balancer in ap-southeast-2.

Global Accelerator vs. CloudFront

Although both services leverage AWS edge locations, their primary goals differ:

FeatureAWS Global AcceleratorAmazon CloudFront
Primary Use CaseOptimize application traffic (TCP/UDP) over AWS global networkCache and deliver static & dynamic web content
IP ManagementStatic anycast IP addresses (or BYOIP)Domain name (CNAME) & edge caches
Traffic OptimizationGlobal anycast routing, health checks, failoverEdge caching, origin shielding
Protocol SupportTCP & UDPHTTP, HTTPS, RTMP
Typical WorkloadsGaming, VoIP, financial applications, IoTWebsites, video streaming, API acceleration

The image is a summary slide discussing CloudFront for caching data at the edge and a global accelerator for routing users to AWS edge locations for optimized network efficiency.

Watch Video

Watch video content

Previous
Cloudfront and LambdaEdge