Your media production company recently moved all its infrastructure into Azure.
Every 14 days you run a batch to render several thousand video clips into various media formats for customers. At the moment the batch job is run on a single H-series virtual machine (VM).
You need to design a scalable compute solution. The solution must meet the following technical and business requirements:
* Must use VM instance sizes smaller than H series
* Must support automatic scale out and scale in based on CPU metrics
* Must minimize deployment time
* Must minimize administrative overhead
What should you do?
You should deploy a virtual machine scale set (VMSS). Scale sets represent the only way to horizontally scale Azure VMs automatically. A scale set is a collection of two or more identically configured Windows Server or Linux VMs that provide full, centralized control over their lifecycle. Scale sets support up to 1,000 instances when you use VM images in the Azure Marketplace.
You cannot configure an auto-scaling rule on the existing VM. Scale sets are the only way to horizontally autoscale Azure VMs. By contrast, Azure App Service apps can be configured individually with auto-scaling rules based on time, date, or CPU metric.
You should not create an Azure Data Factory pipeline. Azure Data Factory is a cloud-based data orchestration engine similar in function to SQL Server Integration Services (SSIS). Therefore, Data Factory is not appropriate for this scenario.
You should not author an ARM template that creates additional VMs. While you can indeed use ARM templates to automate the deployment and removal of VMs, doing so violates the scenario constraints of minimized setup time and management overhead.