← All articles

GCP Committed Use Discounts: A Hands‑On Guide to Real Savings

What are Committed Use Discounts?

Committed Use Discounts (CUDs) are a contract‑style discount that lets you lock in a usage level for a 1‑ or 3‑year term in exchange for a lower price. They apply to many Google Cloud services—Compute Engine CPUs, RAM, GPUs; Cloud SQL vCPU and memory; BigQuery storage; and more. The discount is applied automatically to eligible usage, so you never have to remember to turn anything on or off.

When a CUD Makes Sense – Sizing Your Commitment

1. Identify predictable workloads – Look for services that run continuously or have a steady baseline (e.g., a web‑front tier, a data‑warehouse ETL pipeline, or a long‑running batch job).
2. Gather historical usage – Use the Cloud Billing export to BigQuery or the gcloud billing accounts list + gcloud billing budgets data to calculate average monthly consumption for the past 3‑6 months.
3. Calculate a safe commitment – Aim for 70‑85 % of the average usage. Committing more than you actually consume erodes savings because the excess is billed at on‑demand rates.
4. Prefer 3‑year terms for the biggest discount – GCP offers roughly 30 % off on‑demand prices for a 3‑year commitment versus ~20 % for a 1‑year term, but only if you are confident in the workload stability.

How to Purchase a CUD via Console and gcloud

Console steps

1. Open the Google Cloud Console → Billing → Commitments.
2. Click Create commitment.
3. Choose the service (e.g., Compute Engine), resource type (CPU, RAM, GPU), region, and commitment term (1 or 3 years).
4. Enter the quantity you calculated in the sizing step.
5. Review the price per unit and the estimated annual savings.
6. Confirm the purchase. The commitment appears in the Commitments table and will be applied automatically to matching usage.

gcloud CLI steps

# Example: 4 vCPU, 16 GB RAM in us-central1 for 3 years
REGION=us-central1
CUD_TYPE=COMPUTE_ENGINE
TERM=3Y
VCPU=4
RAM_GB=16

# Create the commitment
gcloud compute commitments create my‑cud \
  --region=$REGION \
  --type=$CUD_TYPE \
  --plan=$TERM \
  --cpu=$VCPU \
  --memory=$RAM_GB

For Cloud SQL or BigQuery storage you would use the beta billing API:

gcloud beta billing accounts commitments create \
  --account-id=YOUR_BILLING_ACCOUNT_ID \
  --service=cloudsql.googleapis.com \
  --resource-type=SQL_INSTANCE \
  --region=$REGION \
  --term=$TERM \
  --amount=2  # number of vCPU‑hours per month

Replace the flags with the values from your sizing exercise.

Managing and Adjusting Commitments

• View existing commitments

gcloud compute commitments list --region=$REGION

• Track utilization – In the console, the Commitments page shows a Utilization % column. In BigQuery, query the gcp_billing_export_v1_ table:

SELECT
  service.description,
  SUM(cost) AS on_demand_cost,
  SUM(commitment_amount) AS committed_amount,
  SUM(cost) / SUM(commitment_amount) AS utilization
FROM `my-billing.project.billing_export`
WHERE service.description = 'Compute Engine' AND usage_start_time >= TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 30 DAY)
GROUP BY service.description;

• Adjust under‑utilized commitments – GCP does not allow you to shrink a commitment, but you can sell the remaining term on the marketplace (if you have a reseller relationship) or simply let it expire and purchase a smaller one for the next term.
• Add additional commitments – If utilization consistently exceeds 85 %, create a new commitment for the excess rather than over‑committing the original one.

Monitoring Savings and Avoiding Pitfalls

• Set up a budget alert tied to the committed‑use discount line item. In Cloud Billing → Budgets, add a filter service.id:COMPUTE_ENGINE and credits.type:COMMITTED_USE_DISCOUNT to see the discount applied each month.
• Watch for regional mismatches – A CUD purchased for us-central1 does not apply to us-east1. If you expand to new regions, you need separate commitments.
• Beware of resource type changes – Switching from n1-standard-4 to e2-standard-4 changes the pricing tier; the original CUD still applies to the generic vCPU and RAM, but you may lose the machine‑type‑specific discount.
• Include GPUs and local SSDs – These have separate CUDs. If you run GPU‑accelerated workloads, purchase a GPU CUD in the same region and term to capture the full discount.
• Review quarterly – Use the Cloud Billing export to BigQuery to run a simple utilization report. If the utilization drops below 60 % for two consecutive months, plan to let the commitment expire rather than renewing at the same level.

How CloudBudgetMaster Automates This

CloudBudgetMaster continuously scans your GCP usage, matches it against active Committed Use Discounts, and surfaces the exact dollar impact of under‑ or over‑utilization. It flags commitments that are below 70 % utilization and recommends the optimal next‑term size, letting you act on savings without manual calculations.