What feature should be utilized for automated scaling based on load for an application hosted on Compute Engine?

The most suitable feature for automated scaling based on load for an application hosted on Compute Engine is Autoscalers, which optimize resource allocation. When an application experiences fluctuations in demand, an autoscaler automatically adjusts the number of virtual machine instances in a managed instance group. This optimizes resource usage by increasing the instance count during peak loads and decreasing it during off-peak times, making it a key component for ensuring efficient resource utilization and cost-effectiveness.

While instance groups do provide the foundation for horizontal scaling, they don’t inherently include the logic for automatic scaling based on load; this is where the autoscaler comes into play. Thus, while instance groups are necessary for enabling multiple instances, it's the autoscaler that actively manages the scaling according to the load.