AWS Wavelength

Why Wavelength? The Ultra-Low Latency Imperative

For a standard mobile application, a request from a user's 5G device typically has to travel from the cell tower, across the mobile network, over the internet to an AWS Region, and then all the way back again. This round trip can introduce significant latency (often >100 milliseconds), which is unacceptable for applications that require near-real-time responsiveness.

Wavelength eliminates this "long haul" journey. By placing compute resources at the edge of the 5G network, application traffic can be processed locally without ever leaving the telecommunications network. This makes it possible to achieve application latencies in the single-digit milliseconds, unlocking a new class of applications.

The Core Architecture of AWS Wavelength

Wavelength provides a seamless extension of your AWS environment, allowing you to use the same APIs, tools, and console you are already familiar with.

Wavelength Zones

A Wavelength Zone is a deployment of AWS infrastructure inside a telecommunication provider’s data center where 5G network traffic is aggregated. It is not a standalone region but rather an extension of a parent AWS Region. This means you can manage your Wavelength resources using the same regional endpoint and console as your other AWS resources.

Extending Your VPC to the Edge

To use a Wavelength Zone, you extend an existing Virtual Private Cloud (VPC) from the parent region into the zone.

1. Opt-in: You first opt-in to the desired Wavelength Zone.
2. Create a Subnet: You create a new subnet within your VPC and explicitly assign it to the Wavelength Zone.
3. Launch Resources: You can then launch AWS resources, such as EC2 instances, directly into this subnet. These resources will physically run in the 5G edge location.

The Carrier Gateway: The Key to Low Latency

The most critical networking component in Wavelength is the Carrier Gateway.

• Purpose: The Carrier Gateway serves two main functions:
  1. It provides connectivity for resources in the Wavelength subnet to the internet.
  2. It allows traffic to flow between the subnet and the telecommunication provider's network, enabling connectivity to mobile devices.
• Low-Latency Path: Crucially, traffic from your EC2 instance destined for the internet or a mobile device routes directly out of the Carrier Gateway at the edge location. It does not have to travel back to the parent AWS Region, which is how the ultra-low latency is achieved.

Resources within the Wavelength Zone can still communicate privately with other resources in your main VPC in the parent region over the secure VPC connection.

Supported Services and Use Cases

Available Services

Wavelength Zones provide a targeted subset of AWS services that are most essential for edge applications:

• Compute: Amazon EC2 (including a selection of general purpose, compute-optimized, and GPU-enabled instances).
• Storage: Amazon EBS (gp2, gp3, and io1 volumes).
• Containers: Amazon ECS and Amazon EKS for running containerized applications.
• Networking: Amazon VPC, Carrier Gateway.

Services that are regional in scope (like S3, DynamoDB, or IAM) are accessed from the Wavelength Zone back over the AWS network to the parent region.

Common Use Cases

Wavelength is built for applications that demand single-digit millisecond latency:

• Interactive Media: Live video and event streaming with interactive elements, removing buffering and lag.
• Connected Vehicles: Real-time communication between vehicles and roadside infrastructure (V2X) for safety and traffic management.
• AR/VR: Offloading heavy rendering and processing from a headset to the network edge to enable high-fidelity, untethered augmented and virtual reality experiences.
• Real-time Gaming: Delivering a responsive, lag-free experience for mobile and cloud-based games.
• Industrial Automation: Processing data from on-site IoT sensors and cameras in near real-time for automated quality control and robotics in smart factories.

Wavelength vs. Local Zones vs. Outposts

AWS offers several services for edge computing. It's important to understand their distinct purposes.