The AWS Global Network is the cornerstone of everything AWS does. It represents an expansive and cutting-edge infrastructure designed to offer unparalleled performance, reliability, and security. It’s more than just a network; it’s a testament to AWS’s commitment to delivering a seamless cloud experience to businesses and end users across the world. By understanding and harnessing the full capabilities of this network, organizations can unlock new potential in application performance, global content delivery, and data transport strategies.

The AWS Global Network offers several near-edge services at the 450+ locations where it peers with the internet. In this chapter, we will cover the following topics:

Overview of the AWS Global Network

Processing at the near edge with Amazon CloudFront

Leveraging IP Anycast with AWS Global Accelerator

Using the AWS Global Network as a private WAN

Overview of the AWS Global Network

Networks like those run by AT&T, Comcast, NTT, Tata, or Zayo are known as Tier-1. This means they can reach the entire internet through settlement-free peering. These are the biggest players on the internet, with the most resources. Nevertheless, even Tier-1 ISPs oversubscribe their networks.

This means they sell more bandwidth to their customers than they can deliver – all at once, anyway. They rely on the fact that most customers don’t hammer their full bandwidth allocation 24 hours a day. This is why your ISP probably gives you less upload than download speed, and why they implement data caps. They also use complex QoS mechanisms to deal with the inevitable periods of congestion that result.

Over the years, they have become good at predicting how much oversubscription they can get away with before losing customers to competitors. For context, an ISP that only oversubscribes 25:1 is considered a good one. Ratios as high as 100:1 are not uncommon.

AWS went online in 2006. In the early days, transit between AWS regions happened over Tier-1 networks. After a few years, it became clear that the needs of a cloud service provider were quite different than those of an ISP. Therefore, AWS built a private backbone between all of its regions and edge locations.

AWS Global Network

In contrast to the oversubscription seen with Tier-1 ISPs, AWS’ backbone is overprovisioned. This means AWS maintains additional capacity above and beyond what it needs most of the time. Rather than rely on QoS to deal with congestion caused by unexpected bursts of traffic, AWS builds to prevent congestion from happening in the first place.

This allows AWS to ensure a deterministic level of performance. It also allows them to do things such as allow MTUs of 9,001 between regions or implement optimizations such as TCP termination from its edge POPs

This total control over all transit is the basis of services such as AWS Global Accelerator, AWS CloudFront, and AWS CloudWAN – all of which will be covered later in this chapter (see the Processing at the near edge with Amazon CloudFront, Leveraging IP Anycast with AWS Global Accelerator, and Using the AWS global backbone as a private WAN sections for more details).