Prepare and test your skills

Prepare and test your skills

Evaluating benefits and tradeoffs across service models means comparing Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS) on scalability, cost efficiency, operational overhead, flexibility, and modernization potential. The central tradeoff is control versus abstraction. More control gives more flexibility but requires more management work, while more abstraction reduces operational overhead but limits customization.
Infrastructure as a Service (IaaS) (opens in a new tab) is the cloud model that gives the customer the most control over the computing environment. In IaaS, the customer manages the operating system (OS), middleware, and applications, while Google manages virtualization, servers, storage, and networking. Google Cloud IaaS examples include Compute Engine (opens in a new tab) and Google Cloud VMware Engine.
IaaS benefits center on flexibility and choice of resources. Compute Engine runs a wide range of workloads with choices of machine series, storage services, accelerators, and guest operating systems, including graphics processing units (GPUs) or tensor processing units (TPUs) on virtual machine (VM) or bare metal instances. Google Cloud VMware Engine migrates and runs VMware workloads natively on Google Cloud without re-architecting existing environments, which supports modernization. The tradeoff is high operational overhead because the customer must patch, configure, and maintain the OS, middleware, and applications, and cost efficiency depends on the resources selected within Google Cloud quotas.
Managed platforms and services act as Platform as a Service (PaaS) by balancing control with ease of use. In this model, Google manages the underlying infrastructure, OS, and runtime environment so the customer can focus on code, containers, and jobs. Examples include Batch, Cloud Run, Cluster Director, Cluster Toolkit, and Google Kubernetes Engine (GKE).
Managed platforms reduce operational overhead by handling provisioning and lifecycle tasks. Batch schedules, queues, and runs batch jobs at scale and manages the lifecycle of the Compute Engine resources it provisions. Cloud Run abstracts away all infrastructure management to run containers that respond to web requests or events, and Cluster Director automates setup of artificial intelligence (AI), machine learning (ML), or high performance computing (HPC) clusters. GKE manages the Kubernetes control plane while the customer deploys container workloads, with Autopilot or Standard modes offering different levels of infrastructure control. The tradeoff is less ability to customize the OS and runtime, in exchange for built-in scalability, high availability backed by service level agreements (SLAs), and resilience such as GKE regional clusters surviving zone outages.
Software as a Service (SaaS) is the model where the provider runs the complete application for the customer. Terraform Cloud is a SaaS application that runs Terraform in a stable, remote environment and securely stores state and secrets. Data flows from the customer's code and version control into the hosted application, and the customer uses the result through the application or its application programming interface (API).
SaaS offers the lowest operational overhead because the provider manages the application environment. Terraform Cloud integrates with the Terraform command-line interface and connects to version control systems like GitHub, GitLab, and Bitbucket, so new commits can automatically trigger Terraform plans. The tradeoff is limited control and flexibility compared with IaaS or PaaS, since the customer depends on the provider for features and the runtime. For organizations that need full change management with a self-hosted option, Terraform Enterprise provides a self-hosted distribution of Terraform Cloud.
The three models differ across the same dimensions of control, work, flexibility, scalability, and modernization. IaaS keeps the most work with the customer, PaaS splits platform work with Google, and SaaS shifts application operation to the provider. Cost efficiency follows the same pattern because IaaS cost depends on configured resources, while PaaS and SaaS reduce manual management effort.
| Model | Control | Operational overhead | Scalability and flexibility |
|---|---|---|---|
| IaaS | Customer manages OS, middleware, and applications; Google manages virtualization, servers, storage, and networking | High | Configurable VM and bare metal instances; choice of machine series, storage, GPUs, TPUs, and guest OSs; VMware migration without re-architecting |
| PaaS / managed platforms | Google manages infrastructure, OS, and runtime; customer manages workloads | Lower; Google manages lifecycle, control plane, or all infrastructure | Cloud Run is highly scalable; Batch runs jobs at scale; GKE regional clusters resist zone outages; Autopilot or Standard controls customization |
| SaaS | Provider runs complete application | Lowest; provider manages application environment | Provider-managed scale; flexibility limited to application features; version control commits can trigger plans |
Exam tip: IaaS requires you to manage the OS, middleware, and applications; managed platforms shift OS and runtime management to Google, while SaaS shifts the entire application to the provider.
Shared responsibility shifts with the service model and directly affects operational overhead. In IaaS, the customer owns the OS, middleware, and applications. In managed platforms, Google also owns the OS and runtime. In SaaS, the provider owns the whole application. Infrastructure as Code (IaC) reduces manual work across all three models by defining infrastructure in versioned configuration files that state the desired state. Those files can be reused to create reproducible development, test, and production environments with source control, continuous integration and continuous deployment (CI/CD) pipelines, and audit history. Google Cloud IaC options include Terraform for human-readable configuration, Infrastructure Manager for automating Terraform deployment, Terraform Cloud for remote SaaS automation, and Terraform Enterprise for self-hosted control.
Three vertical stacks compare IaaS, PaaS and SaaS layers from application to networking, showing customer responsibility shrinking as Google and provider management grows, with Compute Engine, Cloud Run, GKE and Terraform Cloud as examples.
Assessing shared responsibility implications means identifying which security, compliance, maintenance, and operational tasks belong to the customer and which belong to Google Cloud in IaaS, PaaS, and SaaS. IaaS leaves the most responsibility with the customer, PaaS shifts platform management to Google Cloud, and SaaS shifts the most responsibility to the provider. As Google Cloud takes on more work, customer control over lower layers decreases, which shapes governance and risk decisions.
Identity and Access Management (IAM) (opens in a new tab) is Google Cloud's system for controlling who can do what on which resources. IAM connects three parts: a principal that is the identity, a role that is a collection of permissions, and a resource that is the Google Cloud object being accessed. Google Cloud operates the IAM service itself, while the customer manages principals, role grants, and allow policies.
Policy inheritance shapes how access risk spreads. Policies attached to container resources like projects, folders, and organizations flow down to child resources, so one role grant can cover many resources. The customer therefore owns decisions about least privilege and inheritance structure, while Google Cloud owns availability and enforcement of the IAM system. Tools like IAM recommender use machine learning to suggest removal of unused roles and help maintain least privilege.
Infrastructure as Code (IaC) is the practice of provisioning and managing infrastructure with code instead of manual clicks or scripts. The ownership of the IaC toolchain shows how responsibility moves between customer and provider. The customer defines configuration files, while different tools change who runs deployment and storage.
The flow of responsibility moves from customer-managed to provider-managed and back for control. Terraform keeps toolchain work with the customer, Infrastructure Manager shifts deployment work to Google Cloud, Terraform Cloud shifts runtime work to the vendor, and Terraform Enterprise returns runtime work to the customer when compliance requires isolation.
Security duties split between protecting data at rest, data in transit, and access to resources. Google Cloud encrypts Cloud Storage data at rest by default, which is Google Cloud's responsibility, while the customer manages access keys and IAM permissions that control who can reach that data. Transfer Appliance encrypts data at rest and lets customers protect data with keys they manage, so key management stays with the customer.
Compliance checks can move earlier in the workflow without changing the split. Security Command Center's IaC validation checks Terraform plans against organization policies and Security Health Analytics detectors before resources are deployed. That flow runs from developer code to policy check to deployment, so developers catch violations before the control plane creates resources. The customer still owns the correctness of the code and policies.
Governance tools help customers enforce rules and reduce risk regardless of service model. Organization policies and security postures set guardrails that IaC and manual changes must satisfy. IAM recommender identifies excess privilege, while Cloud Identity and Google Workspace manage users and groups that become IAM principals.
Account structure is a governance decision with risk consequences. Customers can combine Cloud Identity and Google Workspace in a single account or use separate accounts for staging and production. Separate accounts isolate blast radius between environments, while a single account simplifies user management. The choice depends on how much isolation compliance and risk management require.
Exam tip: The more managed the service model, the more responsibility Google Cloud takes on, and the less control the customer has over the underlying infrastructure.
Differentiating the core functions of IaaS, PaaS, and SaaS means explaining how each model divides control, management, and abstraction in Google Cloud. Compute (opens in a new tab) in Google Cloud refers to virtualized processing power and memory that runs applications, from raw VM instances to fully managed platforms. Abstraction increases from IaaS to PaaS to SaaS, so operational work falls as provider management rises.
Infrastructure as a Service (IaaS) is the cloud model that offers the most customer control. In IaaS, the customer manages the OS, middleware, and applications, while Google manages virtualization, servers, storage, and networking. This split lets teams build custom environments from scratch without buying physical data centers.
Google Cloud IaaS solutions include Compute Engine and Google Cloud VMware Engine. Compute Engine creates and runs VM and bare metal instances with choices for machine series, storage services, accelerators such as GPUs or TPUs, and guest OSs. Google Cloud VMware Engine runs VMware workloads natively on Google Cloud, where Google manages hardware and the VMware platform while the customer retains full control over VMs, guest OSs, and applications without re-architecting. IaaS best supports workloads needing deep customization and is often preferred by infrastructure architects, ML infrastructure engineers, and platform engineers.
Managed platforms and services function as Platform as a Service (PaaS) by letting customers ship code while Google runs the platform. In this model, Google manages infrastructure, OS, and runtime or control plane, and the customer manages code, containers, jobs, and workload configuration. The workload flows from customer code or containers to Google-managed infrastructure, which provisions resources, runs the job, and manages its lifecycle.
Google Cloud managed platforms include Batch, Cloud Run, Google Kubernetes Engine (GKE), and Cluster Director. Batch schedules, queues, and runs large-scale batch jobs by provisioning Compute Engine resources without requiring the customer to manage VMs directly. Cloud Run is a fully managed serverless platform for highly scalable containerized applications that respond to web requests or events without managing servers or clusters. GKE is a managed Kubernetes environment where Google manages the control plane while the customer deploys container workloads in Autopilot or Standard modes. Cluster Director automates AI, ML, or high performance computing clusters for training, inference, physics simulations, weather models, or molecular dynamics. A specialized PaaS form is Integration Platform as a Service (iPaaS) with Application Integration and Integration Connectors, which connects applications with point-and-click configuration. PaaS best supports containerized apps, batch jobs, and AI workloads where speed matters more than OS control.
Software as a Service (SaaS) is the cloud model with the highest abstraction, where the customer consumes a finished application. In SaaS, the provider runs the software in a stable, remote environment and handles storage, state, secrets, and updates. Data flows from existing customer tools into the hosted application, and the customer signs in and uses features through the application or API.
Terraform Cloud (opens in a new tab) is the SaaS example described in Google Cloud documentation. Terraform Cloud runs Terraform remotely, securely stores state and secrets, integrates with the Terraform command-line interface, and connects to GitHub, GitLab, and Bitbucket so new commits can automatically trigger plans. Terraform Enterprise packages the same function as a self-hosted distribution for an organization's own environment. SaaS best supports tasks where a ready-made capability solves the problem, such as change-managed automation, rather than building servers, runtimes, or platforms.
The three models differ by who manages each layer and what the customer operates day to day. As abstraction rises, customization of lower layers falls. The decision rule is direct: choose IaaS for full OS and stack control, choose PaaS to ship code or large jobs without managing servers, and choose SaaS when a ready application already fits.
| Model | Who manages infrastructure and OS | What customer manages | Google Cloud example |
|---|---|---|---|
| IaaS | Google manages virtualization, servers, storage, and networking | OS, middleware, guest OSs, and applications on VMs | Compute Engine VMs and bare metal; Google Cloud VMware Engine VMs |
| PaaS | Google manages infrastructure, OS, and runtime or control plane | Code, containers, jobs, and workload configuration | Cloud Run containers; GKE workloads; Batch jobs |
| SaaS | Provider manages remote environment, state, and application | Use of application and connection to code and version control | Terraform Cloud hosted automation |
Exam tip: In IaaS you manage the OS and apps while Google manages physical infrastructure; in managed PaaS Google also manages the OS and runtime; in SaaS you only use the hosted application.
Infrastructure as a Service (IaaS) leaves the customer managing the operating system, middleware, and applications while Google manages virtualization, servers, storage, and networking, Platform as a Service (PaaS) shifts operating system and runtime management to Google so the customer manages code, containers, and jobs, and Software as a Service (SaaS) shifts the entire application to the provider so the customer only uses the hosted application. As abstraction increases from IaaS to PaaS to SaaS, operational overhead falls while customization of lower layers also falls.
Choose Infrastructure as a Service (IaaS) such as Compute Engine or Google Cloud VMware Engine when workloads need deep customization with choice of machine series, storage services, GPUs, TPUs, and guest operating systems, or VMware migration without re-architecting. Choose Platform as a Service (PaaS) such as Cloud Run, Batch, or Google Kubernetes Engine (GKE) to ship code or large jobs without managing servers, and choose Software as a Service (SaaS) such as Terraform Cloud when a ready application already fits the task.
In Infrastructure as a Service (IaaS) the customer owns the operating system, middleware, and applications, in managed platforms Google also owns the operating system and runtime, and in Software as a Service (SaaS) the provider owns the whole application. For example, Google Cloud encrypts Cloud Storage data at rest by default while the customer manages access keys and Identity and Access Management permissions that control who can reach that data.
A research organization is migrating its high-performance simulation workloads to Google Cloud. The organization's infrastructure architects have established the following technical and operational requirements:
Which cloud service model and management structure best supports these requirements?
Infrastructure as a Service (IaaS), where Google manages the guest OS and middleware while the organization manages the underlying virtualization and physical servers
Software as a Service (SaaS), where Google manages the entire software stack, application logic, and underlying infrastructure for immediate end-user consumption
Infrastructure as a Service (IaaS), where the organization manages the OS, middleware, and applications while Google manages virtualization, servers, storage, and networking
Platform as a Service (PaaS), where Google manages the underlying infrastructure, OS, and runtime environments so the organization focuses only on running workloads
Last updated