Edit

Bringing and creating Linux images in Azure

  • Article

Caution

This article references CentOS, a Linux distribution that is nearing End Of Life (EOL) status. Please consider your use and plan accordingly. For more information, see the CentOS End Of Life guidance.

Applies to: ✔️ Linux VMs ✔️ Flexible scale sets ✔️ Uniform scale sets

This overview covers the basic concepts around imaging and how to successfully build and use Linux images in Azure. Before you bring a custom image to Azure, you need to be aware of the types and options available to you.

This article will talk through the image decision points and requirements as well as explain key concepts so that you can follow this and be able to create your own custom images to your specification.

Difference between managed disks and images

Azure allows you to bring a VHD to the platform to use as a Managed Disk or as a source for an image.

An Azure managed disk is a single VHD. You can either take an existing VHD and create a managed disk from it, or create an empty managed disk from scratch. You can create VMs from managed disks by attaching the disk to the VM, but you can only use a VHD with one VM. You won't be able to modify any OS properties as Azure will just try to turn on the VM and start up using that disk.

Azure images can be made up of multiple OS disks and data disks. When you use a managed image to create a VM, the platform makes a copy of the image and uses that to create the VM. This allows managed images to support reusing the same image for multiple VMs. Azure also provides advanced management capabilities for images, like global replication and versioning through Azure Compute Gallery (formerly known as Shared Image Gallery).

Generalized and specialized

Azure offers two main image types, generalized and specialized. The terms generalized and specialized are originally Windows terms which migrated in to Azure. These types define how the platform will handle the VM when it turns it on. Both types have advantages, disadvantages, and prerequisites. Before you get started, you need to know what image type you will need. Below summarizes the scenarios and type you would need to choose:

Scenario Image type Storage options
Create an image that can be configured for use by multiple VMs. You can set the hostname, add an admin user, and perform other tasks during first boot. Generalized Azure Compute Gallery or stand-alone managed images
Create an image from a VM snapshot or a backup. Specialized Azure Compute Gallery or a managed disk
Quickly create an image that does not need any configuration for creating multiple VMs. Specialized Azure Compute Gallery

Generalized images

A generalized image is an image that requires setup to be completed on first boot. For example, on first boot you set the hostname, admin user, and other VM-specific configurations. This is useful when you want the image to be reused multiple times and when you want to pass in parameters during creation. If the generalized image contains the Azure agent, the agent will process the parameters and signal back to the platform that the initial configuration has completed. This process is called provisioning.

Provisioning requires that a provisioner is included in the image. There are two provisioners:

These are prerequisites for creating an image.

Specialized images

These are images that are completely configured and don't require VM or special parameters. The platform will just turn the VM on and you will need to handle uniqueness within the VM, like setting a hostname, to avoid DNS conflicts on the same VNET.

Provisioning agents aren't required for these images, however you may want to have extension handling capabilities. You can install the Linux Agent but disable the provisioning option. Even though you don't need a provisioning agent, the image must fulfill prerequisites for Azure Images.

Image storage options

When bringing your Linux image you have two options:

  • Managed images for simple VM creation in a development and test environment.
  • Azure Compute Gallery for creating and sharing images at-scale.

Managed images

Managed images can be used to create multiple VMs, but they have many limitations. Managed images can only be created from a generalized source (VM or VHD). They can only be used to create VMs in the same region and they can't be shared across subscriptions and tenants.

Managed images can be used for development and test environments, where you need a couple of simple generalized images to use within single region and subscription.

Azure Compute Gallery (formerly known as Shared Image Gallery) is recommended for creating, managing, and sharing images at scale. Azure Compute Gallery helps you build structure and organization around your images.

  • Support for both generalized and specialized images.
  • Support for image both generation 1 and 2 images.
  • Global replication of images.
  • Versioning and grouping of images for easier management.
  • Highly available images with Zone Redundant Storage (ZRS) in regions that support Availability Zones. ZRS offers better resilience against zonal failures.
  • Sharing across subscriptions and even between Active Directory (AD) tenants using Azure RBAC.
  • Scaling your deployments with image replicas in each region.

At a high level, you create a gallery and it is made up of:

  • Image Definitions - These are containers that hold groups of images.
  • Image Versions - These are the actual images.

Hyper-V generation

Azure supports Hyper-V Generation 1 (Gen1) and Generation 2 (Gen2). Gen2 is the latest generation and offers additional functionality over Gen1. For example: increased memory, Intel Software Guard Extensions (Intel SGX), and virtualized persistent memory (vPMEM). Generation 2 VMs running on-premises have some features that aren't supported in Azure yet. For more information, see the Features and capabilities section in this article. Create Gen2 images if you require the additional functionality.

If you still need to create your own image, ensure it meets the image prerequisites and upload to Azure. Distribution specific requirements:

Next steps

Learn how to create an Azure Compute Gallery.