The Commitment Coverage Dilemma
Acme Corp runs 400 EC2 instances across three AWS accounts. Their current commitment coverage is 62% via a mix of 1-year no-upfront Reserved Instances (RIs) and Compute Savings Plans (SPs). Monthly on-demand spend is $180,000, and the finance team wants to reach 80% coverage. The CFO is concerned about lock-in because the company plans to migrate 30% of workloads to a new region in 9 months. The engineering team is experimenting with Graviton instances, which are not covered by existing RIs. The FinOps team proposes purchasing 3-year all-upfront SPs to maximize discount. However, the platform team warns that such a commitment could be wasted if the migration happens faster than expected. The company has a 12% cost of capital and a 30% tax rate. They need to decide on the commitment strategy to present at the next quarterly review.
1. Evaluate the proposed 3-year all-upfront SP purchase. What factors should be considered, and what alternative commitment strategy would you recommend? Provide quantitative reasoning where possible.
Read the worked answer
The 3-year all-upfront SP offers the highest discount (up to 50% off on-demand) but carries significant risk due to the planned migration and Graviton adoption. Key factors:
- Lock-in risk: 30% of workloads moving in 9 months means 30% of the commitment could become unused if not offset by new workloads. Unused SPs can be sold on the AWS Marketplace, but typically at a 10-20% loss, and the market is illiquid.
- Technology risk: Graviton instances are not covered by existing RIs but are covered by Compute SPs. However, if the team shifts to Graviton, the SP still applies (since Compute SPs are flexible across instance families), but the discount may be lower than expected if the SP was sized for x86.
- Financial risk: All-upfront requires $180k * 12 * 3 * 0.5 = $3.24M upfront (assuming 50% discount). The cost of capital (12%) means the NPV of the discount is reduced. The tax rate (30%) affects after-tax cost.
Recommendation: Instead of a single 3-year all-upfront SP, adopt a layered approach:
- Purchase 1-year no-upfront Compute SPs for the stable baseline (e.g., 50% of current on-demand spend) to maintain flexibility.
- Use 3-year partial-upfront SPs for the portion of workloads that are certain to remain in the current region (e.g., 40% of on-demand).
- Leave 10% on-demand to accommodate migration and experimentation. This balances savings and flexibility. Quantitatively, a 1-year no-upfront SP might offer 30% discount, while 3-year partial-upfront offers 40%. The blended discount would be around 35%, still achieving significant savings while reducing risk.
2. How should the FinOps team monitor and adjust the commitment strategy post-purchase? Outline a governance process.
Read the worked answer
Post-purchase governance should include:
- Monthly coverage and utilization review: Track SP/RI utilization rates; target >95% utilization. Use Cloudability's commitment utilization reports.
- Expiration calendar: Maintain a 12-month rolling calendar of expiring commitments to avoid coverage gaps.
- Workload change tracking: Integrate with the migration project plan; if migration accelerates, adjust future purchases and consider selling unused commitments.
- Scenario modeling: Use Cloudability's planning tools to model different migration speeds and their impact on coverage.
- Approval gates: Any commitment purchase above a threshold (e.g., $50k) requires FinOps and Finance approval.
- Quarterly business review: Review strategy with CFO, including risk metrics like coverage ratio, utilization, and potential waste.