Answer: Implement a scheduled action that sets the desired capacity to 20 shortly before the office opens.

## Explanation The problem described is a classic **cold start** issue where the application is slow at the beginning of the workday because the Auto Scaling group has scaled down to only 2 instances overnight. When staff arrive and start using the application, the system needs time to scale up from 2 to 20 instances, causing poor performance during the initial surge. Let's analyze each option: **Option A: Implement a scheduled action that sets the desired capacity to 20 shortly before the office opens.** - ✅ **CORRECT**: This proactively scales up the instances before staff arrive, ensuring sufficient capacity is available when the workday begins. - ✅ **Cost-effective**: Instances are only running at full capacity during work hours, not overnight. - ✅ **Predictable**: Since the usage pattern is predictable (office hours), scheduled scaling is appropriate. **Option B: Implement a step scaling action triggered at a lower CPU threshold, and decrease the cooldown period.** - ❌ **Incorrect**: This is reactive scaling that waits for high CPU usage before scaling. By the time CPU thresholds are triggered, users are already experiencing slowness. - ❌ **Not cost-minimizing**: Lower thresholds might cause unnecessary scaling during non-peak times. **Option C: Implement a target tracking action triggered at a lower CPU threshold, and decrease the cooldown period.** - ❌ **Incorrect**: Similar to option B, this is reactive scaling. Target tracking is good for maintaining performance but doesn't solve the cold start problem. - ❌ **Not optimal**: The cooldown period prevents rapid scaling, but decreasing it might cause scaling oscillations. **Option D: Implement a scheduled action that sets the minimum and maximum capacity to 20 shortly before the office opens.** - ❌ **Incorrect**: Setting both minimum and maximum to 20 would prevent any scaling down overnight, increasing costs unnecessarily. - ❌ **Not cost-minimizing**: This would keep 20 instances running 24/7, which defeats the purpose of scaling down overnight. **Key AWS Concepts:** 1. **Scheduled Actions**: Best for predictable workload patterns with known peak times. 2. **Desired Capacity**: The number of instances the Auto Scaling group attempts to maintain. 3. **Cold Start Problem**: When applications need time to scale up from minimal capacity, causing poor initial performance. **Best Practice**: For predictable workloads with known business hours, use scheduled scaling to ensure capacity is available before demand increases, while still allowing scaling down during off-hours to save costs.