How to scale Azure App Service instance size up or down on a schedule?

Unfortunately, there is no way to scale the Azure App Service instance size (i.e. the App Service Plan Pricing Tiers) on a schedule basis at this time.

As of now, Azure App Service can only support horizontal scaling (i.e. instance count scaling) on a schedule basis but not for the vertical scaling (i.e. instance size scaling).

Hope this helps!

Actually you can scale up (vertically, i.e. change the service plan) as well as out (instance count) automatically.

The scale out option has been there forever and allows you to setup rules (e.g. CPU exceeds %, memory goes over threshold, etc.)

The scale up option requires using Azure Automation. It's fully documented here

Hope that helps!

Due to the lack of a simple solution, I created a one-click deploy to do what you're asking.

https://github.com/jraps20/jrap-AzureVerticalScaling

Overview

My approach uses Azure Automation Runbooks. Via the one-click deploy method you can get fully up and running in a few minutes. The two complementary runbooks (ScaleAppServicePlansUp and ScaleAppServicePlansDown) work together to store, read and modify whichever App Service Plans you choose. The primary target for these runbooks are non-prod environments.

The code is too long to include in this answer, unfortunately (so yes, this will mostly be a link-only answer).

Pseudo Code

Scale Down

Iterate across all Resource Groups (or pass in specific one)
Iterate across all App Service Plans (or pass in specific one)
Iterate across all App Services (identify Tier-specific settings)

During iteration, the current App Service Plan Service Tier is stored in Azure Automation Variables (3 distinct variables for each App Service Plan)

Within each App Service Plan, each App Service is iterated to identify tier-specific settings. Some of these settings include: AlwaysOn, Use32BitWorkerProcess, and ClientCertEnabled. All current settings are stored in Azure Automation Variables.

All App Service Plans are then scaled down to the FREE tier.

Scale Up

Iterate across all Resource Groups (or pass in specific one)
Iterate across all App Service Plans (or pass in specific one)
Iterate across all App Services (identify Tier-specific settings)

During iteration, the original App Service Plan Service Tier is retrieved from Azure Automation Variables (3 distinct variables for each App Service Plan)

Within each App Service Plan, each App Service is iterated and any previously stored tier-specific settings are retrieved.

All App Service Plans are then scaled up to their original tier.
All App Services with tier-specific settings are reapplied to their original values.

Additional Resources

• Video tutorial of one-click deployment
• Additional write-up with Sitecore (enterprise CMS on PaaS) example:

Disclaimer

I became aware of Sam Spoerles technique after completing my work. The benefits of my approach over his are as follows:

• One-click deployment to Azure
• Generic or specific based on parameters
• Uses updated Az modules instead of AzureRm modules
• Relies on storage to retain the previous state, via Automation Variables