A solutions architect is investigating compute options for a critical analytics application. The application uses long-running processes to prepare and aggregate data. The processes cannot be interrup

Setting only `Min capacity` and `Desired capacity` means the Auto Scaling group will not scale out to handle occasional usage surges, failing to meet that requirement.

Using Spot Instances is unsuitable for processes that 'cannot be interrupted' because Spot Instances can be terminated by AWS with two minutes of notice, leading to data loss or process restarts.

An Auto Scaling group with `Min capacity` and `Desired capacity` set for the baseline load, `Max capacity` for surges, and a scaling policy allows dynamic scaling for cost-effectiveness. Using On-Demand Instances ensures that critical, uninterruptible processes run reliably without the risk of interruption.

Re-architecting to AWS Lambda functions would likely exceed Lambda's execution duration limits (15 minutes) for 'long-running processes' that prepare and aggregate data, making it an unsuitable and potentially more expensive option for this use case.