First, let’s define some terms that represent concepts that carry specific meanings within the context of AWS Outposts rack:
AWS Outposts Rack: A managed service provided by AWS that is distinct from AWS Outposts server. One deployment of AWS Outposts rack can consist of multiple physical racks.
Rack: When the word rack is used without the AWS Outposts prefix, it refers to a specific 42U 19” preconfigured unit provided by AWS. Each rack houses some combination of compute, storage, and/or networking hardware.
Deployment: One or more racks, all of which operate in concert and share certain logical constructs, such as a service link or Local Gateway (LGW). A deployment can consist of up to 96 racks and is logically tied to a specific Availability Zone (AZ) within its parent region.
Site: The physical location where one or more deployments reside.
Resilience zone: A concept within an Outposts rack that represents a failure boundary for resources, enhancing fault tolerance and availability. A common use of these is to ensure two EC2 instances run within different racks of a deployment.
Use cases
Here are some common customer use cases for all members of the AWS Outposts family:
The need for very low latency and high-throughput connectivity to AWS services from customers’ on-premises infrastructure
The need for analytical tools used for ML training, forecasting, and more to be positioned close to large amounts of data that are unable to move to the cloud because of bandwidth constraints, time constraints, or cost concerns
The need to leverage local AWS capabilities from legacy applications that are poor candidates for migration to the cloud because they contain low-latency integrations with other on-premises systems
Data sovereignty laws that require that its citizens’ data only be accessed from within that country’s borders, making it necessary to bring the needed cloud capabilities into the country applying those regulations
To contrast AWS Outposts rack with on-premises installations, we will walk through the ordering process and the elements you must consider when planning a deployment.