General availability: Azure Monitor predictive autoscale for Azure Virtual Machine Scale Sets
Published date: 13 October, 2022
Predictive autoscale uses machine learning to help manage and scale Azure Virtual Machine Scale Sets with cyclical workload patterns. It forecasts overall CPU load to your virtual machine scale set based on your historical CPU usage patterns. By observing and learning from historical usage, it predicts the overall CPU load ensuring scale-out occurs in time to meet demand.
Several key capabilities are now generally available:
- New virtual machine scale set instances are added when the system predicts that the percentage CPU goes beyond your scale-out boundary.
- You can configure how far in advance you would like new instances to be provisioned.
- You can view predicted CPU forecast without triggering the scaling action with forecast only mode.