Task 4.2: Analyzing and defining business processes.

• AOutsource cloud operations to a managed service provider for the next five years

  Incorrect

  Outsourcing can run systems, but it does not build the company's own skills or bridge its teams.

• BLet each engineering team adopt cloud tools on its own timeline and in its own way

  Incorrect

  Uncoordinated adoption deepens the silos and fragmentation the company already struggles with.

• CForm a cross-functional cloud team of business and technical leads

  Correct

  The Adoption Framework ties adoption to collaboration across business and technical teams and to continuous upskilling of the staff.

• DBuy certification vouchers for the engineers and leave team structures unchanged

  Incorrect

  Certifications help upskilling, but alone they do not improve business and technical communication.

Skills and communication are organizational problems. A cross-functional cloud team, often called a cloud center of excellence, gives business and technical leaders one forum and drives training and shared practices.

• AResource-based commitments for today's machine series in each region

  Incorrect

  Resource-based CUDs apply to specific resources in a specific region, which the company expects to change.

• BSpot VMs for the steady baseline of production workloads

  Incorrect

  Spot VMs can be preempted at any time, which does not suit steady production workloads.

• CSustained use discounts that apply automatically to running VMs

  Incorrect

  Sustained use discounts need no purchase but are smaller and apply only to some machine series.

• DCompute flexible committed use discounts on an hourly spend amount

  Correct

  Compute flexible CUDs are spend-based: the company commits to hourly spend, and the discount applies across eligible machine series and regions.

Cost optimization over a multi-year horizon means committing where usage is predictable while keeping flexibility where it is not. Spend-based flexible CUDs commit to an amount, not to a configuration.

• ABuy three-year committed use discounts for the dev and test VMs

  Incorrect

  A commitment bills around the clock for three years, for machines used about a quarter of the week.

• BAttach instance schedules that stop the VMs outside office hours

  Correct

  Instance schedules start and stop VMs at set times, so the company stops paying for compute while nobody uses it, roughly three quarters of the week.

• CRecreate the dev VMs as Spot VMs that engineers start each morning

  Incorrect

  Spot VMs can be preempted at any time, which would interrupt engineers in the middle of their work.

• DMove the VMs' external traffic to Premium Tier networking

  Incorrect

  The network tier changes how external traffic travels, not what the idle VMs cost, so the compute bill stays the same.

Resource optimization for non-production starts with not paying for idle time. Scheduling VMs to match working hours removes most of the waste without changing how engineers work.

• AA pilot with a group of agents, training, and feedback before a phased rollout

  Correct

  Starting small, training people and acting on their feedback builds the skills and buy-in that a wider rollout needs.

• BA switch of the whole contact center to the new tools on one date, with a manual

  Incorrect

  A single cutover without practice puts customer service at risk and ignores the agents' concerns.

• CUse of the new tools required by policy from launch day and tracked per agent

  Incorrect

  Requiring use before people have practiced or been heard tends to produce resistance rather than adoption.

• DOutsourcing the contact center to a vendor that already uses similar AI tools

  Incorrect

  Outsourcing sidesteps the change rather than managing it, and it is a separate business decision.

Change management is a business process the architect must plan for. Cloud adoption guidance stresses continuous learning, and a pilot-then-scale approach lets people build skills and shape the tools before everyone depends on them.

• ABudgets with alerts plus a Pub/Sub notification to a function that disables billing

  Correct

  Budgets send alerts at thresholds, and programmatic notifications to Pub/Sub can trigger a function that disables billing on the project.

• BBudgets with email alerts at 50%, 90% and 100%, which also halt the spending at 100%

  Incorrect

  Budget alerts notify people; on their own they do not stop resources or spending.

• CQuota limits on each sandbox project that are lowered as the month goes on

  Incorrect

  Quotas cap resource counts, not money, and adjusting them by hand is not automatic.

• DA monthly review of the billing reports, with teams told to delete their resources

  Incorrect

  A monthly review is too late to stop an overspend while it happens.

Cost control for experiments combines visibility and automation: alerts for people, and a programmatic response when a hard limit is reached. Disabling billing stops all services in the project, which suits sandboxes, not production.

• ACosts become fixed monthly fees that stay the same whatever the usage

  Incorrect

  Cloud costs mostly vary with usage; that is why they must be monitored and optimized.

• BSpend stays capital expenditure, because cloud resources are company assets

  Incorrect

  Most cloud consumption is an operating expense; only a few cases, such as sole tenancy, may be treated differently.

• CCosts disappear for operations staff, since Google runs everything

  Incorrect

  Google runs the infrastructure, but the company still operates, secures and optimizes its workloads.

• DSpend moves from upfront hardware purchases to paying for what is used

  Correct

  On-premises IT combines capital purchases depreciated over years with operating costs; in the cloud you provision what you need and pay for what you consume.

Understanding CapEx and OpEx helps stakeholders plan budgets. The cost optimization pillar explains how the cloud's consumption model differs from on-premises.

• AApply the strictest backup policy to every one of the 60 applications

  Incorrect

  Protecting minor systems like critical ones raises costs, because smaller RTO and RPO values cost more.

• BRedesign every application for a multi-region deployment before writing the plan

  Incorrect

  A blanket redesign ignores which applications actually need regional resilience.

• CLet each application team choose a backup schedule that suits it

  Incorrect

  Without business-defined targets, teams cannot know what level of protection is enough.

• DRun a business impact analysis to set RTO and RPO targets for each application tier

  Correct

  DR planning begins with a business impact analysis that defines RTO and RPO, which then drive the right protection for each application.

Business continuity starts from business impact, not technology. Tiering applications by impact lets each one get the recovery targets, and cost, that it warrants.