Prepare and test your skills

Prepare and test your skills

Applying infrastructure models to real-world workload scenarios means matching each workload to private, hybrid, or multicloud based on workload pattern, modernization goal, and integration needs. A private environment keeps a workload in its existing data center, a hybrid architecture connects that private environment to Google Cloud, and a multicloud architecture spreads workloads across more than one cloud provider. Google Cloud supports all three approaches with open-source-based services, managed platforms, and migration tools that reduce interoperability, manageability, cost, and security complexity.
Cloud-first hybrid integration is the pattern where new workloads launch in the public cloud while existing workloads stay where they are. A workload stays in a classic private computing environment only when public cloud deployment is impossible for technical or organizational reasons. Cloud-first is forward looking because new services start in modernized form without requiring everything to migrate at once.
Cloud-first works best for selected workloads that gain the most from the cloud, not for every workload by default. New workloads can be a small fraction of the total environment, so migrating a costly existing workload can sometimes save more than placing a new one in the cloud. A strict cloud-first rollout can also add redundancy, slow performance from excessive cross-environment communication, and conflict with data privacy or industry rules for sensitive data.
A common hybrid example keeps legacy applications holding critical data in the private environment and connects them to new cloud applications through APIs. Apigee (opens in a new tab) provides that API management layer with minimal application changes, while adding security, analytics, and scalability to the legacy services. Data flows from the private legacy system through Apigee to the new cloud services that consume it.
Modernization goals and phased migration paths describe why organizations modernize and the order in which they move workloads to Google Cloud. Modernization and hybrid multicloud adoption reinforce each other because the public cloud makes modernization easier, and modernized workloads get more value from the cloud. The primary modernization goals are:
Migration to Google Cloud uses six related migration types that can be combined for different workloads:
A phased path often starts with lifting and shifting or replatforming, then refactoring or rearchitecting after the workload runs in the cloud. Lifting and shifting first optimizes the workload from an infrastructure perspective, and running in the cloud then makes it easier to adopt cloud services while the workload still communicates with other environments over a hybrid network connection. For example, a large monolithic virtual machine (VM) application can later be broken into independent microservices on Google Kubernetes Engine (GKE) (opens in a new tab) or Cloud Run. If the current architecture cannot run in the target cloud as it is, the path should start with replatforming, refactoring, or rearchitecting to remove that constraint.
Mix-and-match decisions by workload pattern mean different workloads, and even different components of the same application stack, can follow different migration approaches at the same time. The choice is driven by the business and technical requirements of each workload rather than by one organization-wide rule. This approach lets a team modernize where it pays off while leaving stable or constrained components with lighter changes.
One application stack can therefore split across Google Cloud services by component need. The on-premises self-hosted MySQL backend can be replatformed to managed Cloud SQL (opens in a new tab), the frontend VMs can be refactored to containers on GKE Autopilot where Google manages nodes, scaling, and security, and hardware load balancing plus web application firewall (WAF) capacity can be replaced with Cloud Load Balancing (opens in a new tab) and Google Cloud Armor. Traffic flows from users through Cloud Load Balancing and Google Cloud Armor to the container frontend on GKE Autopilot, which then reads and writes data in Cloud SQL.
Three migration approaches differ in how much changes and when each fits the workload pattern:
| Approach | What changes | Choose when the workload pattern shows |
|---|---|---|
| Rehost (lift and shift) | Move with little change | Few environment dependencies, refactoring is not worthwhile before migration, or based on third-party software |
| Refactor (move and improve) | Untangle and improve for cloud | Dependencies must be untangled, platform cannot run in the cloud as is, inefficient use of compute or storage, or cannot deploy automatically |
| Rebuild (remove and replace) | Replace the application | No longer meets requirements, can be merged into another application, technology reached end of life, or license fees are no longer economical |
Exam tip: Choose rehost for small dependencies or third-party software, refactor when dependencies, platforms, or automation block efficient cloud use, and rebuild when requirements, end of life, or license cost make replacement cheaper.
Business and technical fit for hybrid and multicloud is the set of lasting drivers, temporary needs, and technical requirements that together decide workload placement. Lasting multicloud drivers include data sovereignty when a provider has no local region, access to unique capabilities from each provider, avoiding vendor lock-in, and higher resilience by routing traffic to a second provider in the same region if one region fails. Cost reduction requires caution because management complexity, consistent security, software integration, cross-cloud performance, multicloud skills, and separate pricing tools can outweigh service savings.
Temporary hybrid or multicloud fits short-term situations such as limited-time high-scale projects that still need on-premises data, multi-year digital transformation waves, and post-merger integration where one company uses private data centers and the other uses cloud. Technical drivers that support those goals include advanced analytics and AI, faster automated rollouts, high-level APIs, rapid compute and storage provisioning, serverless elasticity, and global infrastructure. A temporary research environment for three to six months, for example, can use on-demand cloud capacity with quick setup to avoid capital expenditure while keeping existing data on premises.
Each workload moves through the same ordered journey, and different workloads can sit in different phases at the same time:
Workload placement should favor measurable business effect such as innovation with cloud AI trained on on-premises data or lower total cost of ownership. Open cloud practices and managed open-source services like Kubernetes help contain the added complexity, and programs such as the Rapid Migration Program and a Cloud Center of Excellence help standardize decisions across teams.
Infrastructure selection is the choice among private, hybrid, and multicloud models based on business needs and rules. Private infrastructure is an on-premises or private data center environment, hybrid connects private or on-premises systems with cloud services, and multicloud uses two or more cloud service providers (CSPs). Data residency, compliance obligations, security requirements, and workload sensitivity are the main factors that shape this choice.
Regulatory and data residency factors are rules about where data must be stored and how it must be protected, and they vary by industry and by region or country. The most common data sovereignty scenario is an organization expanding into a new region or country and having to comply with new data-hosting regulations. If the existing CSP has no local cloud region in that country, the common compliant solution is to use another CSP that has a local cloud region.
When data residency rules require sensitive data such as personally identifiable information (PII) to stay within a specific country or region, a multicloud approach can provide a compliant solution by using two CSPs in one region to provide resilience to outages while meeting the restriction. Data classification for regulated industries helps identify storage locations and cross-region data movement restrictions for certain data classes. Sensitive Data Protection (opens in a new tab) is a fully managed Google Cloud service that discovers, classifies, and protects data assets.
Data movement planning identifies business use cases and the data that powers them, and considers storage type, capacity, accessibility, and movement options. Google Cloud data movement options help enterprises move, integrate, and transform data across environments, which can make hybrid and multicloud adoption easier while meeting specific use cases.
Exam tip: Data sovereignty rules can force a multicloud choice: if the current cloud provider has no local region in a country, using another provider with a local region is the common compliant solution.
Security and workload sensitivity factors determine how much control, visibility, and consistency an organization needs across environments. Hybrid and multicloud architectures can increase the attack surface depending on how systems and data are distributed, raising the risk of unauthorized access, data loss, and other security incidents. On-premises security approaches are not always technically feasible to extend to the cloud.
Each CSP has its own security models, best practices, infrastructure and application security capabilities, compliance obligations, and service names. Shared responsibility models can differ, so organizations must identify the exact demarcation of responsibilities in a multicloud architecture. Observability is key because each cloud typically provides tools to monitor security posture and misconfigurations, but using them separately creates siloed visibility and prevents advanced threat intelligence across the whole environment. A centralized visibility model is recommended to avoid manual correlation between different tools and dashboards.
Recommended security approaches for hybrid and multicloud include:
Google Cloud governance capabilities for these needs include Security Command Center (opens in a new tab), a built-in security and risk management solution that identifies misconfigurations and vulnerabilities, Security Health Analytics, a managed vulnerability assessment scanning feature of Security Command Center that identifies risks and provides remediation recommendations, and Mandiant Attack Surface Management for Google Cloud, which discovers assets from multiple cloud providers, DNS, and the extended external attack surface to help prioritize remediation.
Business drivers and tradeoffs explain why an organization might choose hybrid or multicloud and what costs come with each reason. Each driver supports a business goal but also creates management, security, integration, or cost work that must be weighed before placement. The main drivers and their considerations are:
| Driver | What it supports | Key consideration |
|---|---|---|
| Heeding laws about data sovereignty | Compliance when expanding into a new region | If the current CSP lacks a local region, using another CSP with a local region is the common solution |
| Reducing costs | Lower spending | Account for building and operating across multiple clouds, not only service costs and discounts |
| Using unique capabilities from each CSP | Access to new technologies from more than one provider | Assess feasibility and effectiveness, including common integration challenges |
| Avoiding vendor lock-in | Freedom to choose the best solution | Feasibility depends on dependencies, interoperability, rebuilding costs, skills, and consistent security |
| Enhancing reliability and availability | Resilience to outages | If one CSP region fails, traffic can route to another CSP in the same region if both support the required services there |
Common challenges of multicloud include increasing management complexity, maintaining consistent security, integrating software environments, achieving consistent cross-cloud performance and reliability, building a technical team with multicloud skills, and managing product pricing and management tools from each CSP. Without unified cost visibility and dashboards, managing costs across multiple environments is difficult, and Looker (opens in a new tab) cloud cost management can help where applicable. Long-term benefits matter because deploying on multiple clouds for disaster recovery may increase short-term costs but can prevent outages that cause long-term financial and reputational damage.
Temporary and transformation drivers are short-term or transitional reasons to use hybrid or multicloud rather than as a permanent target. Temporary drivers include reducing capital expenditure (CAPEX) for short-term projects, provisioning infrastructure quickly for limited-time projects that need high-scale distributed infrastructure while still using on-premises data, and supporting a proof of concept with advanced cloud technologies for a certain period. Transformation drivers include supporting multi-year digital transformation projects that align infrastructure and application modernization with business priorities, and integrating IT systems after a corporate merger where companies use different providers or private versus cloud environments.
The most common technical driver for temporary hybrid and temporary multicloud architectures is to facilitate a migration from on-premises to the cloud or to an extra cloud. Cloud migrations almost always naturally lead to a hybrid cloud setup because enterprises transition applications and data systematically based on priorities. A short-term setup therefore often exists while workloads move in waves, with connectivity established first and validation completed before production.
Google Cloud governance and portability considerations are the services and practices that help manage hybrid and multicloud complexity while meeting regulatory requirements. Google Kubernetes Engine (GKE) (opens in a new tab) runs modern applications anywhere at scale and includes managed open source components to secure workloads, enforce compliance policies, and provide deep network observability and troubleshooting. GKE lets organizations operate multi-cluster applications as fleets, and GKE Multi-Cloud supports a consistent governance, operations, and security posture across clouds.
GKE can also support a zero trust distributed architecture, where services deployed on-premises or in another cloud communicate across environments through end-to-end mutual TLS (mTLS) secure service-to-service communications. Workload portability has limits because an application might move with minimal changes but may not perform equally well due to differences in underlying compute, infrastructure security capabilities, networking infrastructure, and proximity to dependent services. Moving a workload may also require moving data, and load balancers provisioned with Kubernetes or GKE may behave differently between environments. Google Cloud's Open cloud approach provides maximum choice and abstracts complexity, supports open source, and helps migrate, build, and optimize applications across hybrid and multicloud while minimizing vendor lock-in and meeting regulatory requirements. A Cloud Center of Excellence (COE) can help adopt cloud faster, drive standardization, and align business strategy with cloud investments.
Choosing among private, hybrid, and multicloud depends on which constraints dominate for each workload. Private infrastructure, such as an on-premises private data center, is the environment an organization may already operate, hybrid connects that private environment with cloud services, and multicloud uses two or more CSPs. Regulatory data residency and data sovereignty can push toward a provider with a local region or toward multicloud when two providers in one region are needed for resilience and compliance.
A decision tree that guides workload placement: workloads stay private when public cloud is impossible, adopt hybrid when legacy on-premises systems must integrate with Google Cloud, and adopt multicloud when data sovereignty, unique capabilities, lock-in avoidance, or resilience across providers dominate, subject to capability parity checks.
Security and workload sensitivity push toward unified security architecture, centralized visibility, standardized design, layered controls, and services such as Security Command Center, Security Health Analytics, Mandiant Attack Surface Management, and Sensitive Data Protection. Business drivers such as cost, unique capabilities, vendor lock-in avoidance, and reliability must be weighed against multicloud challenges including management complexity, consistent security, integration, cross-cloud performance, skills, and cost visibility. Temporary needs such as short-term projects, digital transformation, or merger integration may make hybrid or multicloud a transitional state rather than a permanent target.
Private, hybrid, and multicloud models differ by where workloads and data run, how much control the organization keeps, and how environments connect. Private infrastructure keeps workloads and data in an organization's own environment, such as an on-premises private data center. Hybrid infrastructure connects that private environment with a public cloud, and multicloud infrastructure uses two or more public cloud providers. Google Cloud supports hybrid and multicloud designs with open-source-based services, data movement options, and centralized security and visibility tools.
Private infrastructure is an organization's own computing environment, often an on-premises private data center, where the organization controls the systems and data directly. This model can help with control, security, and compliance because resources stay inside the organization's boundary. Connectivity in a private model stays within the organization's own network and data center, unless the organization connects it to a public cloud to form a hybrid model.
Existing private environments may make it difficult to build advanced capabilities such as analytics and artificial intelligence (AI), and they may not provide the same rapid provisioning of compute and storage as cloud environments. This can limit scalability compared with cloud environments. Private infrastructure often becomes one side of a hybrid architecture when an organization migrates workloads or needs cloud scale while keeping some data on premises.
Hybrid infrastructure combines an on-premises or private environment with a public cloud, so workloads and data can run in both places and connect across a boundary. Organizations often use hybrid architectures as a temporary state or a transition during migration. Common business drivers include reducing capital expenditure (CAPEX) for short-term projects, provisioning infrastructure quickly, supporting limited-time high-scale projects while using on-premises data, managing multi-year digital transformation, and creating a temporary architecture after a corporate merger.
Technical drivers for hybrid infrastructure include building advanced analytics and AI capabilities that might be difficult in existing environments, improving service quality and performance, automating and accelerating application rollouts, using high-level application programming interfaces (APIs) to speed development, accelerating provisioning of compute and storage, using serverless services for elastic scale, and using global infrastructure for multi-regional architectures. The most common technical driver is to facilitate migration from on-premises to the cloud or to an extra cloud, and cloud migrations almost always naturally lead to a hybrid setup.
In a hybrid architecture, services deployed on-premises or in another cloud can communicate across environments through end-to-end mutual Transport Layer Security (mTLS) in a zero trust distributed architecture. Google Kubernetes Engine (GKE) (opens in a new tab) can run modern applications anywhere at scale and operate multi-cluster applications as fleets, with cloud-like experiences on premises and GKE Multi-Cloud for consistent governance, operations, and security posture across clouds. Hybrid architectures add complexity in interoperability, manageability, cost, and security, including a larger attack surface, inconsistent security models, different shared responsibility boundaries, and siloed observability. Data movement is also complex, so teams must identify use cases, data types, storage type, capacity, accessibility, and movement options.
Multicloud infrastructure uses two or more public cloud providers. Organizations adopt it for several business drivers, including data sovereignty when an existing cloud service provider (CSP) has no local region, cost reduction, unique capabilities from each CSP, vendor lock-in avoidance, and reliability and availability. If one region of a CSP goes down, traffic can be routed to another CSP in the same region, assuming both providers support the required capabilities there. Multicloud can also support data residency by using two CSPs in one region to provide resilience while keeping sensitive data such as PII within that location.
Multicloud adoption requires careful tradeoff analysis because service savings can be outweighed by operating costs and constraints. Before adopting multicloud for resilience, teams should assess data movement frequency and possible transfer charges, security and manageability complexities, capability parity between CSPs in the selected region, and whether the technical team has the required multicloud skills. Short-term costs may be higher, but long-term benefits can prevent outages and reputational damage.
Common challenges of multicloud include increasing management complexity, maintaining consistent security, integrating software environments, achieving consistent cross-cloud performance and reliability, building a technical team with multicloud skills, and managing product pricing and management tools from each CSP. Without unified cost visibility and dashboards, managing costs across multiple environments is difficult, and Looker (opens in a new tab) cloud cost management can help where applicable.
Exam tip: Multicloud resilience depends on parity: both providers in the chosen region must support the required services, and data movement, security, and skills must be checked before relying on failover.
Comparing the models shows how private, hybrid, and multicloud split by location, control, connectivity, and operational tradeoffs. The location decides where workloads and data run, control decides who manages systems and boundaries, and connectivity decides how traffic and data move between environments. The following comparison summarizes those differences:
| Model | Where workloads and data run | Control and connectivity | Best fit | Operational implications |
|---|---|---|---|---|
| Private | Organization's own environment, often on-premises | Organization controls systems and data; connectivity stays inside its network unless connected to cloud | Control, security, compliance, or existing on-premises data | May struggle with advanced analytics and AI; may lack rapid provisioning; often becomes one side of hybrid |
| Hybrid | Private environment plus a public cloud | Workloads span both sides; services communicate across environments with end-to-end mTLS in zero trust architecture | Migration, temporary capacity, high-scale projects using on-premises data, transformation, post-merger integration | Complexity in interoperability, cost, and security; larger attack surface; siloed observability; complex data movement |
| Multicloud | Two or more public cloud providers | Workloads distributed across CSPs; connectivity and data movement cross provider boundaries | Data sovereignty, cost reduction, unique capabilities, avoiding lock-in, resilience | Management complexity; hard to keep security consistent; integration and performance challenges; skills and cost visibility issues |
Security, data, and operations across hybrid and multicloud describe the shared risks and practices that apply once systems and data spread beyond one environment. Hybrid and multicloud architectures increase the attack surface as systems and data spread across environments, raising risks of unauthorized access and data loss. It is not always feasible to extend on-premises security approaches to the cloud, because many networking security capabilities are cloud-first and operate in a distributed manner.
Recommended approaches for secure hybrid and multicloud architecture include developing a unified, tailored cloud security strategy, considering a unified security architecture across environments, standardizing design and deployments especially for security, using multiple controls in a layered defense-in-depth approach, monitoring and continuously improving security postures, and considering cloud security posture management (CSPM) for misconfigurations and vulnerability assessment. Centralized visibility avoids manual correlation between different tools, while open practices and managed open-source services help contain complexity across providers and environments.
Private infrastructure keeps workloads and data in an organization's own environment such as an on-premises private data center and best fits control, security, compliance, or existing on-premises data. Hybrid infrastructure combines a private environment plus a public cloud and best fits migration, temporary capacity, high-scale projects using on-premises data, transformation, and post-merger integration. Multicloud infrastructure uses two or more public cloud providers and best fits data sovereignty, cost reduction, unique capabilities from each provider, avoiding vendor lock-in, and resilience.
If the existing cloud service provider has no local cloud region in a country, the common compliant solution is to use another cloud service provider that has a local cloud region. Using two cloud service providers in one region can also provide resilience to outages while keeping sensitive data such as personally identifiable information (PII) within that location to meet data residency restrictions.
Choose rehost, also called lift and shift, when the workload has few environment dependencies, refactoring is not worthwhile before migration, or it is based on third-party software. Choose refactor, also called move and improve, when dependencies must be untangled, the platform cannot run in the cloud as is, compute or storage is used inefficiently, or automatic deployment is not possible. Choose rebuild, also called remove and replace, when the application no longer meets requirements, can be merged into another application, technology reached end of life, or license fees are no longer economical.
A financial services company wants to deploy new, modernized applications in the public cloud while keeping its existing legacy workloads in its on-premises data center. The company has the following business and technical requirements:
Which infrastructure model and Google Cloud approach best fits this scenario?
A public cloud rebuild architecture that decommissions the on-premises systems and rewrites them as cloud-optimized applications in Google Cloud
A multicloud rehost architecture that lifts and shifts the legacy workloads across multiple public cloud providers without modifying the code
A private hosting environment in a colocation facility where the organization maintains control over the virtualization hypervisors for all workloads
A cloud-first hybrid architecture using Apigee to expose on-premises legacy data through APIs to new applications in Google Cloud
Last updated