Variables and Presets

The Key to Reusable Infrastructure Definitions

Introduction

To enable a blueprint to be customized and reused in different environments, we introduce the concepts of variables and presets. Variables allow different environments to consume the same blueprint but behave differently. If the critical design choices in a blueprint are made into variables, the blueprint will become more reusable for a wider audience. When a blueprint is added to an environment, the values of the variables must be provided before the environment can be applied.

If an environment contains multiple blueprints, the variables must be defined in the same blueprint that they referenced in.

Variables

If you wanted to create a database blueprint that is used in production, staging, and development environments, that blueprint would likely describe disks for data storage. The disk size and type for a production environment may be quite different from that of a development environment. If we take the example from the last section and enhance it, we can make the size of the disk a variable:

variables:
  disk_type:
    description: The type of disk to use for the record and log disks.
    type: string
  log_disk_size:
    description: The size of the database log disk.
    type: string
    default: 64 gb
  records_disk_size:
    description: The size of the database records disk.
    type: string
    default: 256 gb

storage:
  disk:
    records:
      size: (( records_disk_size ))
      type: (( disk_type ))
    logs:
      size: (( log_disk_size ))
      type: (( disk_type ))

In this example, variables replace all of the static properties from the previous example. The variable disk_type has no default value and therefore must be set to a value before an environment containing this blueprint can be applied.

Count Variable

When a resource is counted, an automatic variable named count is created.

Mathematical Expressions

Integer variables can be used in simple mathematical expressions, for example you may choose to use the count variable in a counted resource to spread those resources across availability zones:

storage:
  disk:
    records:
      count: 3
      size: 2 TiB
      type: hdd
      zone: (( ((count - 1) % 3) + 1 ))

When the blueprint is processed, disks will appear in different availability zones:

Disk Name	Zone
records	((1 - 1) % 3) + 1 or 1
records-2	((2 - 1) % 3) + 1 or 2
records-3	((3 - 1) % 3) + 1 or 3

Naming by Variable

You can use variables in the blueprint name of an object, either partially or completely:

variables:
  pod:
    description: The root name of the pod this belongs to.
    type: string
    regex: "a-z"

location:
  folder:
    (( pod ))-internal-(( count )):
      region: preferred
      count: 2
    (( pod ))-external:
      region: preferred
  region:
    preferred:
      aws: us-west-2
      azure: westus2

Padding Numbers

Padding allows you to add leading zeros to an integer variable as part of a name like when using our count variable. You can do this by specifying you want to pad the variable and what the length of the integer should be. For example, to ensure two digit numbers in the example above you can specify pad and then the specify the length of the integer,

(( pad(count, 2) )) will ensure a single digit count renders as 01, 02, 03, etc.

(( pod ))-internal-(( pad(count, 2) ))

Value Limiters

Integers

You can specify min and max range checks for variables.

Strings

You can specify a regex (Regular Expression) that the variable value must adhere to.

Presets

Presets are collections of predetermined variable settings that embody best practices configurations. By providing presets, the blueprint author can make it easier for provisioners to add blueprints to their environments. If the author of the example database blueprint wants to provide presets for the variable definitions above, they may look like this (noting they would need to mark the variables as presets):

variables:
  disk_type:
    description: The type of disk to use for the record and log disks.
    type: string
    preset: True
  log_disk_size:
    description: The size of the database log disk.
    type: string
    default: 64 gb
    preset: True
  records_disk_size:
    description: The size of the database records disk.
    type: string
    default: 256 gb
    preset: True
presets:
  deployment_type:
    production:
      disk_type: ssd
      log_disk_size: 64gb
      records_disk_size: 2 TiB
    development:
      disk_type: hdd
      log_disk_size: 16gb
      records_disk_size: 128 GiB

Once a variable is marked as preset, it may not be set independently, although it may have a default value. Preset variables and non-preset variables can be mixed and more than one preset may be used, although a variable cannot be set in multiple presets.

This technique shows how a single blueprint can drive reuse in different use cases. Businesses can empower developers to deploy the same infrastructure as the production site, but at a reduced scale (and cost). A production environment and a development environment can attach the same blueprint and achieve different configurations controlled by variables.

Presets As a Variable

Presets are also available as variables and can be used throughout the blueprint, for example:

location:
  folder:
    (( deployment_type ))-internal:
      region: preferred
  region:
    preferred:
      aws: us-west-2
      azure: westus2

Popular Variables

The following decision points are often made into variables to make blueprints more reusable:

• The region to deploy into.

• The folder resource name to use.

• The number of virtual machines to create.

• The size and type of disks.

• Secrets