It’s no secret that cloud cost overruns are a mounting challenge. Underused machines, dynamic workloads, and shifting project priorities all contribute to wasted spend. At ProsperOps, our mission is to eliminate this inefficiency through automation. We started on AWS Compute in 2018, refining our algorithms and automation before expanding our Autonomous Discount Management (ADM) to AWS RDS, ElastiCache, MemoryDB, Redshift, and OpenSearch. From there, we brought the same detail-oriented approach to Google Cloud Compute and Microsoft Azure Compute, delivering significant savings across these platforms.
Now, we’re taking on one of the biggest cost drivers for organizations using Google Cloud: Cloud SQL. In fact, we found that Cloud SQL was the second largest spend category for our Google Cloud Compute customers.
Introducing ADM for Cloud SQL on Google Cloud
ADM for Cloud SQL optimizes for Spend-based Committed Use Discounts (CUDs), which can range from 25% for a 1-year commitment to 52% for a 3-year commitment and is powered by our proven Adaptive Laddering methodology. We automatically purchase Spend-based CUDs in small, incremental “rungs” over time – rather than a single, batched commitment – to maximize Effective Savings Rate (ESR) and reduce Commitment Lock-In Risk (CLR).
The Cloud SQL Spend Robo-Advisor
Our approach to Cloud SQL rate optimization is similar to how a robo-advisor might ladder fixed income investments in personal finance. Instead of committing a large portion of spend in a single CUD, we spread commitments out in smaller increments. Each commitment expires on a different timeline, allowing our platform to rebalance automatically as your usage changes.
Why it matters:
- You can safely aim for higher coverage (even 100%) because if your usage drops, commitments will expire in staggered intervals.
- You avoid being stuck with a single massive commitment that no longer matches your needs.
To demonstrate the effectiveness of this strategy, we’ll look at covering a 12-month period with the traditional batch purchase approach and with our adaptive laddering approach.
Traditional Batch Purchase
- An organization runs a multi-region PostgreSQL cluster to support a new product.
- After seeing stable usage in one region, they batch-purchase a 1-year CUD covering 83% of their spend.
- Later, testing needs drop, and usage plummets. They’re left paying for an underused, nonrefundable commitment until it expires at the 12-month mark.
ProsperOps’ Adaptive Laddering
- The same organization uses ProsperOps’ ADM for Cloud SQL on Google Cloud.
- Our platform purchases smaller CUD increments over time, with each rung expiring separately.
- When usage drops in a region, the commitments tied to that usage expire and aren’t renewed. When usage picks back up, ProsperOps adds new commitments.
The Result
With ProsperOps’ Adaptive Laddered Approach, you can ensure that commitments have tighter alignment with actual usage, which results in maximized savings while minimizing commitment risk. Our algorithms optimize for ESR and CLR. Assuming the 1-year CUD delivers a 25% discount, comparative ESR and CLR for the 12-month period are as follows:
| Effective Savings Rate | Commitment Lock-in Risk | |
| Traditional Batch Purchase Approach | 13.3% | 12 months |
| Adaptive Ladder Approach | 20.8% | 9.2 months |
Effective Savings Rate increased by 57% while Commitment Lock-In Risk reduced by 23%!
Automatic Optimization with Settings You Control
Cloud SQL commitments support MySQL, PostgreSQL, and SQL Server, and each engine can come with different sizing and regional cost structures. On top of that, Cloud SQL commitments are region-specific, so they do not float across the entire global environment.
Our platform continuously rebalances your commitments to keep pace with real-time changes so you don’t have to. If you reduce usage or shift to another region, ProsperOps automatically adjusts your CUD purchases, protecting you from overspending. Just establish a few basic settings, such as which regions to enable, minimum and maximum coverage targets, and a date, if any, in which you plan to vacate a region, and ProsperOps will take care of the rest.
Early Access and Next Steps
Spend-based CUD Adaptive Laddering for Cloud SQL minimizes costs while offering more flexibility than traditional batch purchased commitments. ADM for Cloud SQL is now available in Early Access for our Google Cloud Compute customers.
Existing ProsperOps customers can view their current Cloud SQL savings performance and request a free Savings Analysis to quantify how our service can cut the cost of your Cloud SQL workloads automatically.
If you aren’t yet a ProsperOps customer and would like to learn more, please reach out to us for a free Savings Analysis. More details about ADM for Cloud SQL can be found on our product page.
Prosper On! 🖖
Clay
