Overview
Pods are running container instances that can be pulled from a container registry such as Docker Hub, GitHub Container Registry, Amazon Elastic Container Registry, or any other compatible registry.
Note: When building an image for submodel.ai on a Mac (Apple Silicon), use the flag
--platform linux/amd64to ensure compatibility. This flag is essential because submodel.ai currently supports only thelinux/amd64architecture.
Understanding Pod Components and Configuration
A Pod is a server container created to access hardware, assigned a dynamically generated identifier. For example, 2s56cp0pof1rmt represents one such instance.
A Pod consists of several key components:
Container Volume: Stores the operating system and temporary data.
This storage is volatile and will be lost if the Pod is halted or rebooted.
Disk Volume: Provides persistent storage, similar to a hard disk, retained for the duration of the Pod's lease.
This storage remains intact even if the Pod is halted or rebooted.
Ubuntu Linux Container: Runs almost any software compatible with Ubuntu.
Assigned vCPU and System RAM: Dedicated resources for the container and its processes.
Optional GPUs Available for specific workloads such as CUDA or AI/ML tasks.
Pre-configured Template: Automates software installation and settings upon Pod creation, enabling quick access to various packages.
Proxy Connection for Web Access: Facilitates connectivity to any open port on the container.
Example:
https://[pod-id]-[port number].tun.submodel.ai, such ashttps://2s56cp0pof1rmt-7860.tun.submodel.ai/.
To get started, check out how to Choose a Pod, then follow the guide on Managing Pods.
Learn More
You can quickly launch a running Pod using a template. For further customization, you can configure the following settings:
GPU Type and quantity
System Disk Size
Start Command
To begin, see how to Choose a Pod and refer to the guide on Managing Pods.
Last updated