ARM Nodes Image Caching

How ARM base image caching works

📘

Orka 2.4.x content

This page has not been updated to reflect the changes introduced in Orka 3.0. Some of the information might be outdated or incorrect. Use 2.4.x to 3.0.0: API Mapping and 2.4.x to 3.0.0: CLI Mapping to figure out the correct endpoints and commands.

🚧

ARM-based nodes only

Intel-based nodes do not need to have the base image locally in order to run a VM.

🚧

Quick navigation

Overview | Free Space Management | Examples

Overview

In order to run an ARM-based VM on a specific node, the base image needs to be transferred to the node first. Once the image is transferred to the node, it is cached, so it can be reused for other VMs. This means that the initial deployment on a specific node takes longer, while all other deployments are significantly faster.

Free Space Management

Cached based images use disk space on the ARM-based node. As this space is limited, there might not be enough space to cache a new base image. If a new image needs to be cached and there is not enough space, the old base images are deleted until the new image can be cached.

Orka looks for free disk space that is twice as much as the image. This is needed because of the way VMs are using the base image. This means that if 90GB image is to be cached, the node needs to have at least 180GB free space. If not, the old base images are deleted until there is enough space.

Examples

In a scenario with a host with 1TB disk space, there is roughly 900GB free.
Here are some examples showing how many images can be cached on such a host:

  • 5 90GB images can be cached on the host
  • 2 200GB images can be cached on the host
  • 1 290GB image can be cached on the host
  • 2 90GB images and 1 200GB image can be cached on the host
  • Image bigger than 450GB cannot be deployed at all

© 2019-2023 Copyright MacStadium, Inc. – Documentation built with readme.com. Orka is a registered trademark of MacStadium, Inc.