A public advertisement causes sudden, unpredictable traffic spikes to an application running in an Auto Scaling group. Instance boot includes a lengthy process that creates machine-specific caches, so startup is slow. The Auto Scaling group must be able to handle surges that may require scaling to 100 instances. Which solution ensures capacity is available quickly for these surges?

A warm pool pre-provisions instances in a stopped or running state, making them ready to serve traffic almost immediately when needed. This directly addresses the problem of slow instance startup due to lengthy cache creation, ensuring capacity is available quickly for sudden traffic spikes. Launching 100 instances and stopping them is not a warm pool; it's a manual, less efficient approach that doesn't integrate with Auto Scaling's lifecycle management. Increasing the instance warm-up period only delays scaling actions; it doesn't speed up instance readiness. Predictive scaling uses machine learning to forecast future traffic and scale proactively, but it might not react fast enough to sudden, unpredictable spikes and doesn't solve the slow boot time issue directly.