Skip to main content

Plan a Savings Plan or reservation purchase

Your on-demand compute bill keeps growing, and someone suggests buying a Savings Plan. Before you commit for one or three years, this guide checks that you use what you already own, finds the usage still paying on-demand prices, and points you to a purchase sized for it.

You will use: Commitments · Cost Savings Time: 20 minutes

1. Check that you use what you own​

Open Financial Planning → Commitments and select Commitment Utilization by Type. Set the time range to Last 30 days, grouped Daily.

Commitment Utilization by Type over 30 days, daily: utilization 96.8%. The Compute Savings Plan drops to about 80% on the weekends of 3, 10 and 17 October; the Reservation stays at 100%.

If a commitment type sits well below 100%, you already pay for more than you use. At Northwind Labs (a fictional company), the Compute Savings Plan drops to about 80% every weekend. Buying more would add to that waste: fix utilization first, for example by moving workloads onto the commitment, or wait for usage to grow.

Unused Commitment by Type shows the same gap in dollars.

2. Find the usage that still pays on-demand​

Open Commitment Coverage by Service, Last 90 days, grouped Weekly. A service with low coverage and steady usage is the candidate for a new commitment.

Commitment Coverage by Service, 90 days, weekly: AmazonEC2 reaches 100% in the week of 21 September. Breakdown: AmazonRDS 100%, Virtual Machines 93.2%, Compute Engine 75.3%, AmazonEC2 54.5%; Redshift, Azure SQL Database and Cloud Run 0%.

Look for usage that stays level week after week. A short spike, such as a load test, is not a reason to buy: it would leave the commitment unused when the spike ends.

3. Get a sized recommendation​

Open Cost Management → Cost Savings, select Filter, and narrow it to the Commitment category.

  • AWS: Cost Optimization Hub recommends Compute Savings Plans and RDS Reserved Instances, with an estimated monthly saving and a runbook. Other commitment types appear under Other.
  • Microsoft Azure: Azure's reservation and savings plan recommendations, with Azure's estimate.
  • Google Cloud: Google Recommender's committed use discount recommendations, from your BigQuery export.

Compare each recommendation's amount with the steady usage from step 2. Buy in the provider's console; CloudYali does not purchase commitments.

4. Check the result​

A week or two after the purchase, open Commitment Coverage by Cloud and Commitment Utilization by Type again. Coverage should rise, and utilization of the new commitment should stay near 100%. Then Effective Savings Rate by Cloud shows the saving across the whole bill.

What you have now​

  • Proof that existing commitments are used before you buy more.
  • The services still paying on-demand prices, and how steady their usage is.
  • A purchase sized from the provider's recommendation, and a way to check it worked.

Do this every quarter​

Review utilization and coverage each quarter, and before any commitment expires. CloudYali does not list individual commitments with their expiry dates yet; check those in the provider's console.

Last updated on