Prepare and test your skills

Prepare and test your skills

Cost governance and accountability tools in Google Cloud organize who pays for resources, assign costs to owners, and keep spending within plan. The main structural pieces are the resource hierarchy (opens in a new tab), Cloud Billing accounts (opens in a new tab), labels and tags (opens in a new tab), Identity and Access Management (IAM) roles, budgets (opens in a new tab), alerts, and quotas (opens in a new tab). These pieces work together so spending is visible, responsibility is clear, and controls prevent surprise charges.
The cost governance process moves in a fixed order from visibility to control to reduction. The order matters because reduction only works after spending is visible and new spending is controlled:
The Google Cloud resource hierarchy maps a company's operational layout to cloud resources and controls visibility. A domain manages the users in the organization and relates directly to the organization resource, which represents the whole company. Below the organization sit folders, which can separate departments, teams, or production from development, and below folders sit projects, which contain service-level resources such as virtual machines, storage, and networking.
A hierarchy diagram showing how the domain, organization, folders, and projects relate in Google Cloud, with projects linked to a Cloud Billing account that pays for them and connects to a Google Payments profile.
A Cloud Billing account defines who pays for a set of Google Cloud resources. Each project must be linked to a Cloud Billing account to function, and one billing account can pay for many projects. The billing account tracks all costs, including charges and savings, and links to a Google Payments profile that holds payment methods, address, and tax information. There are two billing account types: self-serve online accounts charge automatically to the linked payment method, while invoiced offline accounts receive a monthly invoice and typically require eligibility criteria.
Access to billing is controlled by IAM roles that separate ownership from viewing and linking. The creator of the billing account automatically receives the Billing Account Administrator role, which has full control. The Billing Account Viewer role can view costs and transactions, while the Billing Account User role can link projects to the billing account.
Cost allocation assigns cloud costs to the teams, projects, departments, or individuals that caused them. Labels and tags are key-value pairs attached to resources, such as team=banking_appdev or region=emea, and they flow into billing exports and reports so costs can be filtered by those dimensions. Chargeback recovers cloud costs from internal customers, which aligns incentives with performance and promotes fiscal discipline.
A useful metadata strategy for cost allocation has validity, automation, and simplicity. Valid tags identify business KPIs and regulatory needs, automated tagging enforces compliance when resources are created, and simple rules avoid confusion for operational teams. This approach makes each owner responsible for their own cloud usage.
Budgets track actual Google Cloud spend against a planned amount. A budget can apply to an entire billing account, specific projects, services, or labels. Budget alerts send notifications when spending crosses thresholds, such as 50%, 90%, and 100% of the budget amount, so stakeholders can act before costs grow further. Budgets by themselves do not automatically cap spending; they provide warnings.
Anomaly detection finds spikes or deviations that differ from expected spending patterns. The Anomalies dashboard shows cost anomalies for projects in the linked billing account, and users can investigate root causes and set thresholds for cost impact and percent deviation. Programmatic notifications can forward budget or anomaly alerts to other channels or adjust quotas, while spend cap budgets can automatically pause eligible services until the cap is manually lifted. The FinOps hub provides personalized recommendations to optimize costs and track FinOps health.
Quotas restrict how much of a shared Google Cloud resource can be used. A team can set its own quota limits up to a cap, for example by limiting how many Compute Engine instances can exist in a given region. This limit blocks resource provisioning beyond the limit and prevents unexpected charges from uncontrolled growth.
Other provisioning controls enforce policy before resources are created. The Organization Policy Service enforces constraints such as restricting certain products or resource creation. Infrastructure as code manages provisioning like application code and enforces policy compliance in continuous integration. Reviewing changes before applying them, documenting each environment's preferred state, detecting drift with Cloud Asset Inventory, and using least-privilege IAM together keep resources aligned with plan.
Cloud Billing reports are the primary tool for understanding what drives costs. These reports show cost trends, filter by project, service, SKU, and labels, and show the impact of credits. The Cost Table provides a detailed tabular view for reconciliation, while Cost Breakdown shows costs after savings so users see the net effect of credits and discounts.
Exporting billing data to BigQuery allows detailed analysis and custom dashboards. Custom dashboards and reports visualize cost data, monitor key performance indicators (KPIs), and filter by time period or resource type. Regularly reviewing tags, labels, and cost tools keeps cost information current and helps identify unexpected increases.
The main financial controls have different jobs. A budget tracks and warns, a quota limits resource use, and a spend cap budget pauses eligible services.
| Control | What it does | Effect on spending |
|---|---|---|
| Budget | Tracks planned versus actual spend for a billing account, projects, services, or labels | Warns; does not automatically cap spend |
| Budget alert | Sends notifications at thresholds such as 50%, 90%, and 100% | Warns stakeholders so they can act |
| Quota | Limits how much of a shared resource can be used, up to a cap | Blocks resource use beyond the limit |
| Spend cap budget | Pauses eligible services when enforced | Stops charges for the service in the project until manually lifted |
Exam tip: A budget warns you about spending, a quota limits resource usage, and a spend cap budget pauses eligible services until you manually lift it.
Capital expenditure (CapEx) and operating expenditure (OpEx) are the two ways an organization accounts for IT spending. CapEx is a large, upfront purchase of a long-lived asset that the organization owns and depreciates over time, such as buying servers and building a data center. OpEx is ongoing, pay-as-you-go spending on services consumed as needed, such as paying a cloud provider for resources a workload actually uses.
Capital expenditure is money spent upfront to acquire a durable asset that the organization owns and uses for years. A traditional data center is the classic CapEx example: the organization buys servers, storage, and networking gear, houses them in a facility, and pays for power, cooling, and staff to run them. Because the purchase happens before any workload runs, the organization must estimate future demand in advance and buy enough capacity to cover peak usage.
That early commitment creates idle capacity and fixed costs. Capacity that is bought for peak periods sits unused during quieter periods, but the money is already spent whether or not it is used. The asset's cost is spread across its useful life through depreciation rather than charged to a single period.
Operating expenditure is ongoing spending on services that are consumed as they are used, with no large upfront purchase. In Google Cloud, the billing model charges for resource usage, so a customer pays for the compute, storage, and network capacity that workloads actually consume. Costs are recorded as they are incurred, which means spending tracks demand instead of being locked in ahead of time.
Because the provider owns and operates the underlying infrastructure, the customer avoids the capital outlay, the depreciation schedule, and much of the maintenance burden. Managed services push this further by shifting operational work such as patching, scaling, and reliability to the provider. This shift reduces the customer's operational overhead and toil.
Consumption-based pricing changes both how quickly an organization can act and how hard it is to forecast spending. On the agility side, teams can provision resources through tools such as the Google Cloud console, the Google Cloud SDK, Cloud APIs, and Terraform, so new capacity is available without a procurement cycle or capital request. Autoscaling reinforces this by adjusting capacity to match demand: Compute Engine managed instance groups add and remove VMs based on load, and Cloud Run scales instances up when traffic rises and down when it falls, which avoids overprovisioning and its unnecessary costs.
On the predictability side, variable usage makes costs less fixed than a depreciation schedule. Spending rises and falls with consumption, so the organization must actively monitor and govern spending rather than rely on a known annual figure. This tradeoff is central: agility increases, but financial predictability requires active controls.
| Dimension | Capital expenditure (CapEx) | Operating expenditure (OpEx) |
|---|---|---|
| What is paid for | A long-lived asset the organization owns | Services consumed as they are used |
| When payment occurs | Upfront, before the workload runs | Continuously, as usage happens |
| Capacity planning | Must estimate peak demand in advance | Capacity can follow actual demand |
| Cost behavior | Fixed and depreciated over time | Variable and tied to consumption |
| Effect of idle capacity | Money is already spent | No charge when resources are not used |
Managing OpEx requires visibility because OpEx varies with usage, and Google Cloud provides tools that make consumption spending visible and controllable. Cloud Billing (opens in a new tab) tracks current and projected spending, and its billing reports show cost trends that can be filtered by project, service, SKU, and labels. Budgets and budget alerts (opens in a new tab) compare actual charges against a target amount and notify stakeholders at thresholds such as 50%, 90%, and 100% of the budget, and budgets can be scoped to an entire billing account, specific projects, services, or labels.
Labels and tags (opens in a new tab) are key-value pairs attached to resources that let an organization break costs down by team, project, or department and support chargebacks and audits. Discounts reduce the rate paid for consumption and trade some flexibility for lower cost. Committed use discounts (CUDs) (opens in a new tab) exchange a commitment to a minimum level of usage or spend over a one-year or three-year term for discounted prices and suit stable workloads, while sustained use discounts reward consistent Compute Engine usage within a billing cycle. Spot VMs (opens in a new tab) offer unused VM capacity at more than 80% cheaper than regular VMs for fault-tolerant and flexible workloads, and a total cost of ownership view should also count engineering effort for patching, scaling, security hardening, and disaster recovery that managed services absorb.
Exam tip: Budgets and budget alerts warn about spending but do not automatically cap it, so they prevent surprises rather than block charges.
Evaluate Pricing Models and Discount Structures means comparing how Google Cloud charges for resource use and how discounts lower the billing rate for different workload patterns. Google Cloud bills for actual resource consumption, so costs rise and fall with usage instead of requiring fixed hardware purchases. Three cost-optimization mechanisms divide the savings by commitment and flexibility: Committed use discounts (CUDs) (opens in a new tab), sustained use discounts (opens in a new tab), and Spot VMs (opens in a new tab).
Committed use discounts are discounted prices offered in exchange for a commitment to use a minimum level of Google Cloud resources for a specified term. The standard commitment term is typically one year to three years, and the discount rewards predictable and steady usage. Because a central FinOps team can see requirements across the organization, it can aggregate demand to maximize commitment-based savings, and CUDs can be used on top of enterprise discounts for significant additional savings.
Google Cloud offers different commitment scopes for different buying needs:
Resource-based CUDs can be shared across all projects associated with most Cloud Billing accounts, so unused commitments in one project can benefit another project under the same account. CUD auto-renewal avoids a gap in savings by automatically repurchasing commitments when they expire. Pricing models can change over time, so commitments and alternative features should be reviewed regularly to keep architecture and billing aligned.
Sustained use discounts are automatic discount credits applied to certain Google Cloud products such as Compute Engine and Google Kubernetes Engine (GKE) after continuous resource usage beyond specific duration thresholds. Sustained use discounts require no upfront commitment, and the discount is earned simply by running the same resource consistently during the billing cycle. This structure benefits workloads that run steadily but were not covered by a prior commitment, and the service tracks continuous usage and applies the credit without action by the user.
Spot VMs are Compute Engine virtual machines that provide access to unused Google Cloud capacity at a lower cost compared to regular VMs. Spot VMs are more than 80% cheaper than regular VMs, which makes them attractive for controlling compute costs. The tradeoff is availability, because Compute Engine might preemptively stop or delete Spot VMs to reclaim capacity when that capacity is needed elsewhere.
Spot VMs are therefore suitable for fault-tolerant and flexible workloads that can tolerate interruption. Batch jobs that can restart and do not have high availability requirements are a typical fit, while workloads that need guaranteed capacity should use a different option.
Choosing the most cost-effective strategy means matching the discount mechanism to how predictable the workload is and whether interruption is acceptable. A steady production database needs guaranteed capacity, while a short batch analysis can restart if capacity is reclaimed. Regular review connects the choice to changing prices and features, because a new machine type, region price, or BigQuery edition might lower cost further.
| Option | How savings are earned | Choose when |
|---|---|---|
| Committed use discounts (CUDs) | Commit to minimum quantity or spend for one to three years | Usage is predictable and steady over a long period |
| Sustained use discounts | Run Compute Engine or GKE resources continuously beyond duration thresholds with no commitment | Usage is consistent during the billing cycle but not pre-committed |
| Spot VMs | Use unused VM capacity at more than 80% lower cost with risk of preemption | Workload is fault-tolerant, flexible, and can tolerate being stopped |
Cost Management (opens in a new tab) tools support this decision after it is made. Budgets and alerts in those tools track actual spending against planned spending and signal when a strategy needs adjustment. Assessing workloads (opens in a new tab) before committing helps confirm whether usage is steady enough for CUDs, consistent enough for sustained use discounts, or flexible enough for Spot VMs.
Exam tip: Choose Committed use discounts (CUDs) for predictable steady usage over one to three years, while Spot VMs fit only fault-tolerant workloads that can tolerate preemption.
Capital expenditure (CapEx) is upfront spending to acquire a long-lived asset the organization owns, such as servers for a data center, with costs fixed and depreciated over time. Operating expenditure (OpEx) is ongoing pay-as-you-go spending for services consumed as they are used, such as Google Cloud compute, storage, and network capacity, so spending varies with consumption.
A budget tracks planned versus actual spend and sends budget alerts at thresholds such as 50%, 90%, and 100% but does not automatically cap spending. A quota limits how much of a shared resource can be used and blocks use beyond the limit, while a spend cap budget pauses eligible services until manually lifted.
Choose Committed use discounts (CUDs) when usage is predictable and steady because they exchange a commitment to a minimum quantity or spend for one to three years for discounted prices. Choose sustained use discounts for consistent Compute Engine or Google Kubernetes Engine (GKE) usage during the billing cycle with no commitment, and choose Spot VMs only for fault-tolerant and flexible workloads that can tolerate preemption because they use unused capacity at more than 80% lower cost with risk of being stopped.
A digital media company needs to run large-scale data processing jobs on Compute Engine. The engineering team has identified the following workload requirements and financial constraints:
Which Google Cloud pricing mechanism or compute option should the company select?
Spot VMs
Resource-based Committed Use Discounts (CUDs)
Sustained Use Discounts (SUDs)
Spend-based Committed Use Discounts (CUDs)
Last updated