Inference
Marketplace
Introduction
The Applications and delivery section in Edge Infrastructure Services is where you run workloads on nodes. It covers four interconnected areas: virtual networks that provide network connectivity for workloads, application instances that run the workloads themselves, Docker Compose support for multi-container applications, and deployment automation for fleet-scale rollouts. Understanding how these areas relate to each other helps you plan your deployment workflow before working through individual tasks.Prerequisites
- You have an active Edge Infrastructure Services enterprise.
- You have the SysManager or SysAdmin role in your enterprise. See Manage Roles.
- At least one node is onboarded and online.
- At least one app image exists in the Build section.
How It Works
The four areas have a dependency order. You do not always need all four. When you do use multiple areas, the sequence that follows reflects the order in which they depend on each other.Virtual networks
Virtual networks provide network connectivity for application instances running on the same node. You create a virtual network, assign it to a node, and configure whether it routes traffic (Local type) or bridges directly to the physical network (Switch type). If your applications need custom networking (multi-path routing with failover, VLAN segmentation, or L2 bridging), create the virtual networks before deploying application instances. Applications that do not require custom networking can use the default network configuration without this step. See Virtual Networks Overview for network types, port options, and IP configuration.Application instances
An application instance is a single workload running on a node: a virtual machine, a container, or a Docker Compose stack. You deploy an app image from the Build section to a target node to create the instance. During deployment, you attach the instance to one or more virtual networks, configure its resources, and set its startup behavior. See Applications Overview for instance concepts, run states, and deployment targets.Docker Compose
For multi-container workloads defined with acompose.yaml file, Edge Infrastructure Services provides a dedicated Docker Compose runtime environment. The Compose Runtime is itself an application instance: a QCOW2 image that runs the Docker engine and a compose agent (the process that manages container lifecycles on behalf of Edge Infrastructure Services) inside an isolated virtual machine. You deploy the Compose Runtime to a node first, then deploy Docker Compose application instances to it.
See Docker Compose Overview for the architecture, the two-layer deployment model, and how compose files are delivered to the runtime.