The Retail Node Components project is one ingredient of Intel's Open Retail Initiative, which is focused on driving workload consolidation and performance at the Edge. This project aims to solve industry challenges by accelerating solution development through automated provisioning and deployment of devices that seamlessly integrate into customer environments and provide workload orchestration, scalability, and clustering.
Figures 1 and 2 illustrate a production level deployment process. A customer places an order with a System Integrator for devices they need for their business use case. The System Integrator specifies the profile associated with the customer's use case in the Retail Node Installer's (RNI) configuration file. This profile defines the entire software stack that the Retail Node Installer will install on the devices. The Retail Node Installer pulls the images and files specified by the profile from the cloud and caches them locally. The System Integrator can then automatically provision the devices by simply plugging them into the Retail Node Installer's Local Area Network and turning them on. Once the devices finish building, they will automatically harden and turn off. The devices are now ready to ship to the customer. At the customer's on-site location, there is no additional setup of firmware, software, tool suites, etc. The profile has already specified every component and service that the customer needs on the devices. The devices are use-case ready and will seamlessly integrate into the customer's environment.
Figure 3 illustrates a sample reference solution for a device. In this example, the profile has defined a software stack of Rancher* OS kernel, Docker*, and some application containers, which have been installed on the device by the Retail Node Installer.
Figure 4 has a more complex software stack with software components like clustering, remote management, and Docker Swarm. When the device joins the user's node cluster, it will seamlessly integrate into the ecosystem, enabling scalability and workload orchestration. We aim to provide a reference design for a profile that integrates these components and services in a future release.