Prepare and test your skills
Prepare and test your skills
License portability with Bring Your Own License (BYOL) is the practice of reusing software licenses an organization already owns on Compute Engine (opens in a new tab) instead of buying new on-demand licenses. The approach is financially attractive when the organization has already paid for physical-core or physical-processor licenses, because Google Cloud adds no extra charge for bringing the image itself. The customer still pays the software vendor under existing agreements and pays Google Cloud for the infrastructure that hosts the licensed virtual machines (VMs).
Sole-tenant nodes are physical servers dedicated to hosting VMs for only one project. VMs from the owning project run on one side of this hardware boundary, while VMs from other customers run on separate physical servers on the other side. Nodes can be configured for different workload needs, including configurations that minimize the number of physical servers and physical cores that must be licensed.
Licenses with dedicated hardware requirements can only run on compliant dedicated hardware in Compute Engine. The customer must bring its own media, run the imported image on sole-tenant nodes in a supported region, and review licensing terms before importing. This hardware boundary is what keeps the existing license portable without breaking dedicated-host rules.
| License scenario | Requires sole-tenant nodes | Portability path |
|---|---|---|
| Physical-core or physical-processor license with dedicated hardware requirement | Yes | Bring own media and run imported image on sole-tenant nodes in a supported region |
| Linux bring-your-own-subscription (BYOS) for RHEL or SLES | No | Bring subscription without sole-tenant nodes |
| Microsoft application such as SharePoint Server or SQL Server with Software Assurance | No, when the operating system allows it | Use Microsoft License Mobility on default multi-tenant instances |
Exam tip: Hardware-bound physical-core licenses require sole-tenant nodes, while RHEL/SLES BYOS and Microsoft License Mobility for eligible applications do not.
The BYOL workflow is the ordered sequence that moves licensed software from on-premises media to running VMs on dedicated hardware and then to vendor reporting. The workflow depends on sole-tenant node templates and node groups to reserve dedicated capacity before any VM can be placed on it.
Prerequisites for BYOL on dedicated hardware determine both license compliance and how much infrastructure must be paid for. Choosing a smaller core footprint and a maintenance policy that keeps VMs within the licensed servers directly reduces the number of physical cores that need coverage. Authentication verifies identity for access to Google Cloud services and APIs before importing and provisioning.
restart-in-place or migrate-within-node-group, the two node group policies that minimize the number of physical servers used for licensing purposes.n1-node-96-624 node type contains 96 CPUs, so a node group with the minimum of 2 nodes requires at least 192 CPUs of quota, plus possible adjustments for VMs and IP addresses.Microsoft license portability on Compute Engine depends on whether the customer holds mobility rights or must rely on dedicated hardware. If the customer brings its own licenses to Compute Engine, the customer is responsible for ensuring compliance with Microsoft under its licensing terms. Questions about license eligibility go to the licensing reseller, while questions about the import and hosting process go to Google Cloud Support.
With Software Assurance with License Mobility, an eligible SQL Server license can run on default multi-tenant instances as long as the underlying operating system does not itself require sole-tenant nodes. The customer completes verification through Google with the License Mobility verification form or directly with Microsoft licensing (opens in a new tab) by naming Google LLC as the Authorized Mobility Partner. Without Software Assurance with License Mobility, SQL Server can only be brought to sole-tenant nodes and only when the license meets the criteria for Outsourcing Software Management Rights, because Microsoft does not allow that right in a multi-tenant environment.
Per-core rules make the financial impact explicit. Windows Server per-core licensing is only available on sole-tenant nodes, where required core licenses equal the number of physical cores on the licensed server, subject to a minimum of 8 core licenses per physical processor and 16 per server, with core licenses sold in 2-packs. A Windows Server Standard license grants the right to host two instances on the physical host, while the Datacenter edition with per-core licensing grants the right to host an unlimited number of instances. For SQL Server on sole-tenant nodes under Outsourcing Software Management Rights, a minimum of four core licenses is required for each physical processor, supported models include per-core, per-processor, and per-server or per-client access license (CAL), and upgrades are not allowed to versions released after October 1, 2019.
Tracking and reporting close the BYOL lifecycle by connecting technical usage to license payments. Google Cloud provides tools for importing images, managing VMs on dedicated hardware with sole-tenant nodes, minimizing physical core usage, and tracking physical core and processor usage for reporting. Data flows from sole-tenant node usage logs to Cloud Logging and then to BigQuery for analysis of physical core counts. The customer then reports that consumption to the vendor and continues to pay for licenses under existing agreements.
The same responsibility principle applies to other enterprise software. Oracle software on Compute Engine follows a BYOL model where the customer is responsible for procuring licenses and complying with Oracle license terms. Reducing capital expenditure therefore comes not from avoiding license fees, but from rightsizing dedicated nodes, minimizing physical servers and cores, and reusing licenses already owned while paying only for the Google Cloud capacity actually used.
Comparative analysis of license-included and marketplace models compares paying for software as part of the Google Cloud hourly price versus bringing an existing vendor agreement to the cloud. License-included images bundle the license into Compute Engine (opens in a new tab) usage, while BYOL lets an organization reuse licenses already purchased from vendors such as Microsoft, Red Hat, SUSE, SAP, and Oracle. The choice affects long-term cost, how easily workloads scale, and how much operational work is needed for compliance.
License-included images and marketplace licenses are options where Google Cloud collects the software charge together with infrastructure charges. With this model there is no separate vendor payment to manage, VM instances can run on standard multi-tenant hosts, and cost grows and shrinks directly with the number of running instances. Oracle Database@Google Cloud is a marketplace example where an organization can purchase the license as part of the Google Cloud Marketplace order instead of bringing its own Oracle license, and networking charges for data transfer between applications and Oracle Exadata databases within the same region are included in the price of that offering. This model fits variable workloads because elastic scaling, rightsizing, and autoscaling directly lower both infrastructure and license cost without tracking physical hardware.
Bring your own license (BYOL) and bring your own subscription (BYOS) are models where the organization brings its own media and images and continues to pay the software vendor under existing agreements. Although there is no additional Google Cloud charge for bringing images with existing licenses, the organization must still pay for its licenses according to vendor agreements and must review licensing terms before migration. Common examples include Linux BYOS with Red Hat Enterprise Linux (RHEL) or SUSE Linux Enterprise Server (SLES), Microsoft applications such as SharePoint Server and SQL Server through Microsoft License Mobility, SAP software which runs on Google Cloud only under BYOL and is not sold by Google Cloud, and Oracle products where the customer is responsible for procuring licenses and complying with Oracle terms.
Licensing scenarios for Linux BYOS with RHEL or SLES and for Microsoft applications do not require dedicated hardware. In contrast, existing physical-core or physical-processor licenses that have dedicated hardware requirements must run on compliant hardware configurations such as sole-tenant nodes.
Exam tip: Physical-core or physical-processor licenses with dedicated hardware requirements require sole-tenant nodes, while RHEL or SLES BYOS and Microsoft License Mobility for SharePoint Server and SQL Server do not.
Sole-tenant nodes are physical servers dedicated to hosting VM instances for only one project, and they are the compliant hardware target for BYOL licenses with dedicated hardware requirements. An organization can bring images with existing licenses in any region that supports sole-tenant nodes, and it can configure node templates and groups to minimize the number of physical servers and cores that must be licensed. Because the license count is tied to physical cores and processors rather than virtual machines, Google Cloud provides tools for importing images, managing VMs on dedicated hardware, minimizing physical core usage, and tracking usage through Cloud Logging and BigQuery.
To bring images with existing licenses to Compute Engine, complete the workflow in order:
Before starting, review the node group host maintenance policy and choose restart-in-place or migrate-within-node-group, because both options minimize the number of physical servers used for licensing purposes. New projects receive a CPU quota of 72, so quota must be increased for node groups; for example, an n1-node-96-624 node type contains 96 CPUs and a minimum group of 2 nodes requires at least 192 CPUs of quota. Enable the Cloud Logging API when agreements require tracking of physical server usage.
Long-term financial impact and scalability tradeoffs compare predictable license reuse against elastic pay-as-you-go growth plus operational work. Total cost of ownership (TCO) includes hardware and software plus operational costs of running a data center such as power, cooling, maintenance, and support services, and it is typically easier to reduce costs with elastic scalability on Google Cloud than with a rigid on-premises data center. A commonly overlooked cloud cost is network utilization, including egress traffic that leaves Google Cloud and regional and zonal traffic that crosses boundaries for high availability and disaster recovery, which can be significant for data-intensive workloads. To estimate Google Cloud resource costs, use the Google Cloud Pricing Calculator and Cloud Billing (opens in a new tab) budgets, reports, and billing export to BigQuery.
The models differ in how cost behaves as environments grow:
| Model | How software is paid | Hardware and scaling effect |
|---|---|---|
| License-included images | License cost is bundled into hourly Compute Engine usage | Runs on shared hosts and scales elastically with autoscaling groups, custom machine types, and rightsizing |
| Marketplace integrated license | License is purchased as part of the Marketplace order, such as Oracle Database@Google Cloud | Managed offering can include same-region data transfer; scaling follows the service rather than physical core tracking |
| BYOL on sole-tenant nodes | License is paid separately to the vendor under existing agreements | Tied to dedicated physical servers and cores; scaling requires adding nodes, quota, and license coverage plus usage tracking and vendor reporting |
Cost control practices apply to all models but matter most when usage scales. Useful actions include evaluating committed use discounts (CUDs), using Spot VMs for fault-tolerant workloads at more than 80% lower cost than regular VMs, decommissioning idle resources and idle VMs, rightsizing overprovisioned instances with machine-type recommendations, configuring automatic scaling for managed instance groups, and migrating from self-managed services to Google-managed services to reduce patching, scaling, hardening, and disaster-recovery work.
Resource optimization and license compliance auditing is the combined practice of sizing cloud resources to actual need and continuously verifying that software use matches vendor agreements. Over-provisioned VMs, disks, and clusters increase infrastructure spend and can also increase licensed capacity, while unmanaged provisioning creates drift that breaks compliance. In Google Cloud, rightsizing, autoscaling, and governance controls work together to prevent over-provisioning and to keep an auditable record of what is running.
Rightsizing is the process of matching provisioned capacity to measured utilization so resources are neither idle nor oversized. Compute Engine (opens in a new tab) supports rightsizing during migration with Migrate for Compute Engine and after VMs are running by applying machine type recommendations for instances. Google Cloud Observability provides monitoring, logging, and performance data that shows how much CPU, memory, and other resources a workload actually uses, which reveals room to downsize. Because migration batches are grouped in part by compliance and licensing requirements, rightsizing decisions directly affect the TCO model built with hardware, software, operational costs such as power and cooling, network charges, and the Google Cloud Pricing Calculator.
Practical rightsizing combines several cost-reduction actions that also reduce license exposure:
When existing capacity no longer fits demand, evaluating whether to grow the current environment or provision a new one keeps growth sustainable and prevents paying for capacity that is not needed.
Autoscaling automatically adds and removes capacity based on workload fluctuations so applications have resources when needed without permanent over-provisioning. Compute Engine managed instance groups (MIGs), Google Kubernetes Engine (GKE), and Cloud Run each provide autoscaling, but they operate at different layers of the stack. Traffic that flows from users to these scaled resources stays within the configured minimum and maximum, so scaling decisions control both performance and spend.
| Option | What is scaled | How scaling is triggered |
|---|---|---|
| Compute Engine managed instance groups (MIGs) | Group of VMs managed as a single entity | Autoscaling policy with minimum and maximum VM counts, for example adding VMs when CPU utilization reaches a threshold and removing them when it drops |
| Google Kubernetes Engine (GKE) autoscaling | Nodes, Pod replicas, and Pod resource requests | Cluster Autoscaler adds or removes nodes based on Pod demands, Horizontal Pod Autoscaler changes replicas based on CPU, memory, or custom metrics, Vertical Pod Autoscaler tunes requests and limits, and Node Auto-Provisioning creates optimized node pools |
| Cloud Run | Serverless container instances | Built-in autoscaling based on incoming traffic, scaling up as traffic increases and scaling down as traffic decreases |
Choosing among them depends on where the workload runs: choose MIGs for VM-based services, choose GKE autoscaling for containerized workloads that need coordinated node and Pod control, and choose Cloud Run when code should run without managing infrastructure.
Software license compliance means running only software for which the organization holds valid rights under the vendor agreement. SAP on Google Cloud (opens in a new tab) runs under a BYOL model, where the customer must obtain SAP software licenses from SAP or an SAP reseller and ensure compliance with SAP terms. Google Cloud does not sell or provide licenses for SAP software, although SAP provides trial or developer edition licenses for non-production use through the SAP Cloud Appliance Library for rapid deployment.
The Google Cloud Rapid Assessment and Migration Program (RAMP) provides a free assessment that counts SAP systems and records CPU type and core count, memory amount, operating system, and storage amount to estimate the Google Cloud infrastructure needed. That inventory links resource sizing to license planning, because core counts, memory, and system counts determine both infrastructure shape and license scope. Compliance and licensing requirements are also explicit criteria for grouping workloads into migration batches and ordering those batches, alongside business value, dependencies, refactoring effort, and availability needs.
Exam tip: SAP on Google Cloud is BYOL — Google Cloud provides SAP-certified infrastructure, not the SAP license itself.
Continuous auditing is an ongoing loop that monitors spend, controls provisioning, and reduces waste to maintain compliance over time. Cloud Billing is the collection of tools for tracking current and projected spend, and cost governance follows an ordered process where each stage depends on the previous one:
Labels and tags are key-value pairs attached to Google Cloud resources that support chargebacks, audits, and cost allocation analysis on top of Cloud Billing. Costs can be broken down by label, while tags support allocation analysis through standard usage exports, and labels also provide extra context in Cloud Logging. Cloud Asset Inventory analyzes assets across projects and services so the current state can be compared to the documented preferred state, unmanaged resources can be decommissioned, and drift can be corrected. Preventive controls enforce boundaries before noncompliant resources are created: the Organization Policy Service enforces constraints on products or resource creation, quotas cap shared resource use such as Compute Engine instances in a region, IAM with least-privilege grants prevents privilege escalation, and centrally aggregated audit logs preserve evidence for review.
Physical-core or physical-processor licenses with dedicated hardware requirements require sole-tenant nodes on Compute Engine, while Linux bring-your-own-subscription for Red Hat Enterprise Linux (RHEL) or SUSE Linux Enterprise Server (SLES) and Microsoft License Mobility for eligible applications such as SharePoint Server or SQL Server do not. To use the dedicated hardware path, you must bring your own media and run the imported image on sole-tenant nodes in a supported region.
License-included images bundle the software cost into hourly Compute Engine usage and run on standard multi-tenant hosts so cost scales elastically with running instances, while Bring Your Own License (BYOL) reuses licenses already purchased with no additional Google Cloud charge for bringing the image. With BYOL you continue to pay the software vendor under existing agreements and pay Google Cloud for infrastructure, and when the license has dedicated hardware requirements cost is tied to dedicated physical servers and cores on sole-tenant nodes.
The workflow is to prepare images according to license agreements, activate licenses as required by the vendor, import virtual disk files and create Compute Engine images, create a sole-tenant node template and then a node group to reserve physical servers, provision virtual machines on the node groups using the imported images, track license and physical core and processor usage with Google Cloud reporting tools, and report consumption to the vendor.
An enterprise is planning to migrate 150 steady-state on-premises Windows Server and Red Hat Enterprise Linux (RHEL) workloads to Google Cloud Compute Engine. The financial operations (FinOps) team is conducting a Total Cost of Ownership (TCO) analysis to compare license-included (Pay-As-You-Go) images against custom-licensed (Bring Your Own License) options.
Which architectural and financial consideration correctly guides their migration strategy?
Last updated