This domain is about 18% of the exam. Scenarios describe an application or data centre and ask how to move or modernize it, or which compute option fits. Learn the migration strategy names and the compute ladder: VMs, containers, serverless.
Migration terms
| Term | Meaning |
|---|---|
| Workload | An application or set of resources doing a job, the unit you migrate |
| Discovery and assessment | Inventorying what you have, its dependencies, cost and fit for the cloud. The first step. Google offers Migration Center for this |
| Retire | Switch off applications no longer needed |
| Retain | Leave as is for now (compliance, recent investment, not ready) |
| Rehost (lift and shift) | Move as is to VMs in the cloud with minimal change. Fastest, least benefit |
| Replatform (move and improve) | Small optimisations during the move, such as swapping a self-managed database for a managed one or moving to containers |
| Refactor | Rewrite or re-architect the application to be cloud-native (microservices, serverless). Most effort, most benefit |
| Reimagine | Rethink the business process and build something new with cloud capabilities, data and AI, rather than moving the old thing |
Order by effort: rehost < replatform < refactor < reimagine. Order by speed: the opposite. “Move quickly with minimum change” means rehost. “Get cloud benefits but not rewrite everything” means replatform. “Cloud-native, maximum agility” means refactor.
Tools: Migration Center (assessment), Migrate to Virtual Machines (VM rehost), Database Migration Service, Storage Transfer Service, Transfer Appliance (physical device for huge data).
Compute terms
| Term | Meaning |
|---|---|
| Virtual machine (VM) | A software-defined computer on shared hardware with its own OS. Full control, you maintain it |
| Container | A lightweight package of an application and its dependencies that shares the host OS kernel. Starts fast, runs the same anywhere, uses fewer resources than a VM |
| Microservices | An application built as small independent services that can be developed, deployed and scaled separately |
| Serverless | You write code or run containers; the provider manages servers, scaling (including to zero) and you pay for actual use |
| Spot VMs | Deeply discounted VMs (60 to 91% off) that Google can reclaim at any time. For fault-tolerant, interruptible work such as batch jobs |
| Kubernetes | Open source system for deploying, scaling and managing containers (orchestration) |
| Autoscaling | Automatically adding or removing resources with demand |
| Load balancing | Spreading traffic across resources for performance and availability |
| Managed services | Services where Google handles setup, patching and scaling |
The ladder from most control to least operational work: VMs (Compute Engine) -> containers (GKE) -> serverless (Cloud Run). More abstraction means less to manage but less control.
Compute Engine
Compute Engine is Google Cloud’s IaaS virtual machine service.
Business value:
- Run almost any workload, including lift and shift of existing applications, with full control of OS and software.
- Wide choice of machine types: general purpose, compute-optimised, memory-optimised, GPU and TPU accelerated; custom machine types.
- Managed instance groups with autoscaling and load balancing.
- Pay per second; discounts through sustained use (automatic), committed use discounts, and Spot VMs.
- Runs on Google’s global infrastructure with live migration of VMs during maintenance (less downtime).
Pick it when you need control of the OS, run software that is not container-friendly, or want the fastest migration path.
Modern application development
Business value of building modern, cloud-native applications:
| Value | How |
|---|---|
| Flexible architectures | Microservices let teams release independently and use the best technology for each part |
| Faster deployment | Managed services and CI/CD pipelines shorten release cycles |
| Cost optimisation | Pay for use; scale to zero; less idle capacity |
| Scalability and resilience | Scale individual services; failures are contained |
| Operational efficiency | Less infrastructure to run, more automation |
A monolith (one large codebase) is simple to start but is slow to change and must be scaled as a whole. DevOps practice (development and operations working together with automation) supports this.
Google Kubernetes Engine (GKE)
GKE is Google’s managed Kubernetes service for containers (Google created Kubernetes).
Business value:
- Portability: containers run the same on premises, on Google Cloud or elsewhere.
- Automated operations: auto-scaling, auto-repair, auto-upgrade of nodes and clusters.
- Efficient use of resources by packing many containers per machine.
- Autopilot mode: Google manages the nodes as well; you only manage workloads and pay for what your pods request. Standard mode gives more control over nodes.
- Strong fit for microservices and for modernizing applications gradually.
Pick GKE when you want container orchestration with control, many services and portability. Pick Cloud Run when you want the simplest path with no cluster to think about.
Serverless: Cloud Run and Cloud Run functions
| Product | What it does | Pick it when |
|---|---|---|
| Cloud Run | Runs containers (any language) as a fully managed serverless service; scales automatically, including to zero | Web apps, APIs, services packaged as containers; no cluster management; spiky or unpredictable traffic |
| Cloud Run functions | Runs small event-driven functions of code (formerly Cloud Functions) | Short single-purpose tasks triggered by events: a file upload, a Pub/Sub message, an HTTP call |
| App Engine | Older PaaS for applications | Mentioned less now; Cloud Run is the modern default |
Serverless benefits: no infrastructure to manage, automatic scaling, pay only when code runs, faster time to market. Trade-offs: less control, and some runtime limits.
Quick map:
- Existing app needing OS control -> Compute Engine.
- Containerised microservices at scale with orchestration -> GKE.
- Container or API, zero ops -> Cloud Run.
- Small event handler -> Cloud Run functions.
Hybrid and multicloud products
Google Cloud emphasises running across environments. Know which products work outside Google Cloud:
| Product | What it does |
|---|---|
| GKE Enterprise | Manage and secure Kubernetes clusters consistently across Google Cloud, on premises and other clouds, with fleet management, policy and service mesh (the successor to Anthos) |
| AlloyDB Omni | A downloadable edition of AlloyDB that runs in your data centre or other clouds |
| BigQuery Omni | Run BigQuery analytics on data stored in AWS or Azure without moving it |
| Cloud SQL | Managed relational databases; also usable in hybrid designs with connectivity |
| Looker | BI that connects to data across clouds and on premises |
| Google Distributed Cloud | Google Cloud infrastructure and services running in your data centre, at the edge, or in air-gapped environments for sovereignty needs |
Business value: modernise without a risky all-at-once move, meet data residency rules, avoid lock-in, use one consistent management layer.
APIs and Apigee
An application programming interface (API) is a defined way for software to talk to other software: a contract describing requests and responses. It hides internal complexity, so a partner app can get data or trigger actions without knowing how your systems work.
Business opportunity from APIs
- Expose capabilities to partners, developers and customers to form ecosystems.
- Monetise APIs: charge per call, per plan or revenue share (a bank offering payments, a weather data seller).
- Connect legacy systems to modern apps by wrapping them with APIs.
- Speed up integration and partnerships.
Apigee is Google Cloud’s API management platform. It sits between API consumers and your back ends to:
- Secure APIs (authentication, quotas, threat protection).
- Manage them (versions, rate limits, policies, developer portals).
- Analyse usage and performance.
- Monetise them with plans and billing.
- Work across hybrid and multicloud back ends.
Pick Apigee when the scenario is about publishing, governing and monetising APIs or modernising legacy systems behind APIs. It can also manage APIs that serve AI agents.
Exam checklist
- Migration strategies: retire, retain, rehost, replatform, refactor, reimagine; first step is discovery and assessment.
- Rehost = fastest, minimum change. Refactor = rewrite, most cloud benefit.
- VM = full control; container = portable and lightweight; serverless = no servers to manage.
- Compute Engine = IaaS VMs. GKE = managed Kubernetes (Autopilot manages nodes too). Cloud Run = serverless containers. Cloud Run functions = event-driven code.
- Spot VMs are cheap but can be reclaimed: only for fault-tolerant jobs.
- Hybrid and multicloud: GKE Enterprise, AlloyDB Omni, BigQuery Omni, Looker, Google Distributed Cloud.
- APIs let you open services to partners and monetise them; Apigee manages, secures and monetises APIs.