FinOps Lab Thursday: AWS Database Savings Plans — Up to 35% Off Your Data Layer
CloudCostChefs TeamAWS Database Savings Plans: A New Tool for FinOps
AWS re:Invent 2025 delivered a gift for FinOps practitioners: Database Savings Plans. Let's break down this new cost optimization tool.
What Are Database Savings Plans?
The same commitment-based discount model you know from Compute Savings Plans, now applied to database services. You commit to a consistent amount of database usage (measured in $/hour) for a 1-year term with no upfront payment:
Covered AWS Database Services
Database Savings Plans automatically apply to eligible usage across these AWS database services:
Available in all AWS Regions except China Regions.
Discount Rates by Service Type
Different database deployment models receive different discount rates:
| Service Type | Discount Rate | Examples |
|---|---|---|
| Serverless | Up to 35% | Aurora Serverless v2 |
| Provisioned Instances | Up to 20% | RDS, Aurora provisioned, ElastiCache |
| DynamoDB On-Demand | Up to 18% | DynamoDB on-demand throughput |
| DynamoDB Provisioned | Up to 12% | DynamoDB provisioned capacity, Keyspaces |
The Real Flexibility Story
Database Savings Plans automatically apply to usage regardless of engine, instance family, size, deployment option, or region. This means:
Change instance types: Move from db.r7g to db.r8g instances and keep your discount
Shift regions: Migrate workloads from EU (Ireland) to US (Ohio) without losing savings
Modernize engines: Move from RDS for Oracle to Aurora PostgreSQL or from RDS to DynamoDB while maintaining discounts
Important Limitations
Before you commit, understand what Database Savings Plans don't cover:
- Storage costs - Only compute/capacity charges are covered, not storage or backup costs
- Older instance generations - Plans exclude previous-generation instance types
- EC2 instances - Cannot move spending between compute and database services
- 3-year terms - Only 1-year commitments available (no longer-term discount option)
Why This Matters
Database costs are often the "hidden ingredient" in cloud bills. They're:
- Sticky (hard to turn off without application changes)
- Growing (data accumulates faster than you plan)
- Overlooked (compute gets all the attention)
For many organizations, databases represent 20-30% of their AWS bill. A 35% discount on that chunk is significant.
Lab Experiment: Calculating Your Savings
Let's run the numbers:
| Scenario | Monthly DB Spend | 35% Savings | Annual Savings |
|---|---|---|---|
| Small | $10,000 | $3,500 | $42,000 |
| Medium | $50,000 | $17,500 | $210,000 |
| Large | $200,000 | $70,000 | $840,000 |
How to Purchase Database Savings Plans
AWS makes it simple to evaluate and purchase Database Savings Plans through two methods:
1Recommendations View
AWS Cost Management Console analyzes your historical usage and recommends optimal commitment levels. Navigate to AWS Billing and Cost Management → Savings Plans → Recommendations.
2Purchase Analyzer
For custom analysis, use the Purchase Analyzer to model different commitment scenarios and see projected savings before you commit.
3CLI & API
Automate purchases using the AWS CLI or API for infrastructure-as-code workflows and programmatic management.
Chef's Implementation Recipe
1Audit Current Usage
Pull 3-6 months of database usage data. Look for:
- • Consistent baseline usage
- • Seasonal patterns
- • Growth trends
2Identify Commitment Level
The savings plan should cover your baseline usage — the minimum you know you'll need. Don't commit to peak usage.
3Calculate Break-Even
With a 35% discount on committed usage, you break even if you use ~65% of your commitment. Below that, you might be better staying on-demand.
4Consider Flexibility Needs
Savings Plans offer more flexibility than Reserved Instances. If you might change database engines or instance types, Savings Plans are likely better.
When to Use vs. Reserved Instances
| Use Savings Plans When | Use Reserved Instances When |
|---|---|
| You want flexibility across services | You have stable, predictable workloads |
| You're unsure about future instance types | You can commit to specific configurations |
| You value simplicity | Maximum discount is the priority |
The Catch
1-year commitment means you're locked in. If your database needs change dramatically, you're still paying.
Chef's Pro Tip
Layer your commitments:
- Use Savings Plans for 60-70% of expected baseline
- Keep 30-40% on-demand for flexibility
- Re-evaluate quarterly as usage patterns become clearer
The Bottom Line
Database costs have been harder to optimize than compute. This changes that. If you have consistent database workloads, Database Savings Plans should be on your Q1 FinOps agenda.
Time to cook up some savings on your data layer.