Container Groups

Container Groups are the basic unit of deployments within SCE. A Container Group is comprised of a container image, a set of hardware requirements (CPU, GPU, RAM...), and a replica count. The Container Group lets SCE the know on how to deploy your container to the Salad network. As the owner of the Container Group it is your responsibility to Start, Stop, or Delete the Container Group, while SCE is responsible for running the Container Group and maintaining the desired replica count.

Replica Count

Container Groups support a replica count, allowing you to define the desired number of Container Instances. Each Container Instance is an exact copy of the Container Image specified in the Container Group, allowing you to scale up (or down) the number of Container Instances running your container. The Replica Count is the only mutable property on a Container Group, allowing you to dynamically scale the number of Container Instances up or down, depending on your specific application needs. SCE currently supports replica counts varying from 1-100 Container Instances per Container Group, please reach out if you need a higher replica count.

Hardware Requirements

When creating a Container Group you are required to specify the hardware requirements to run your container. Salad will only run your application on hardware that meets these requirement.

Per Container Hardware Requirements

  • Number of vCPUs (1-16)
  • Memory (RAM) (1-16)
  • Network Speed
  • Geolocation

Container Networking

Containers run with residential IP addresses and use CloudFlare DNS (, External connections can be “tunneled” to the running container, but ports of the host computer are not exposed.

Container Group Actions

Please see API Reference for more details


Creates a new Container Group, when the Container Group is created it is not yet running on nodes.


Starts a deployment of the container group on the Salad network.


Only the Replica Count may be updated once the Container Group has been created, allowing you to dynamically scale the number of Container Instances running.


Stops a current deployment, stopping all active Container Instances that are running. Container Groups that have been stopped may be started again.


Deletes the Container Group and any information associated with it. This action can not be undone.

Example Workflows

- I need 20 copies of the same container

Create and Start a single Container Group with a replica count set to 20.

- I need to run 10 different containers

Create and Start a 10 unique Container Groups with a replica count of 1.

- I have updated my container and want to rollout the latest version

Create a new Container Group with the latest version of your container. Then slowly increase the replica count of the new Container Group while simultaneously decreasing the replica count of the older version.