A biotech firm runs steady-state genomics APIs on EC2 using 60% average CPU all year plus nightly batch jobs that tolerate interruptions and finish in 3 hours. Budgets are tight and availability matters for APIs. Which purchase option mix based on usage patterns minimizes cost while preserving availability?

The correct mix aligns commitment to usage shape because year-round APIs at 60 percent CPU justify Compute Savings Plans that cut cost while preserving availability across families and regions, while nightly three-hour interruption-tolerant batch fits EC2 Spot Instances that offer deep discounts with reclaim handling and checkpointing. The Convertible Reserved Instances paired with On-Demand batch preserves availability but fails cost efficiency because Convertible flexibility for changing needs is unnecessary for stable APIs and offers roughly the same discount as Compute Savings Plans with less flexibility while On-Demand ignores stated interruption tolerance. The Savings Plans paired with On-Demand batch correctly commits the baseline but fails cost minimization for variable work because it pays full price for jobs explicitly described as interruptible instead of using Spot. The inverted Spot for APIs with Reserved coverage for batch fails both availability and commitment fit because Spot cannot guarantee continuous serving for availability-sensitive APIs and locking brief nightly jobs into continuous commitment wastes spend during idle hours. Exam caveat is to map steady versus bursty versus interruptible patterns to Savings Plans and Reserved versus Spot rather than choosing by discount alone, since interruption tolerance and duration determine suitability. Practical check is to verify baseline with Cost Explorer utilization, confirm batch checkpointing on Spot interruption notices, and model one-year Savings Plans coverage only for the API baseline before purchase.