Google Cloud Digital Leader Practice Worksheets & Answer Key

Answer: Measure how quickly a team can provision an experiment, evaluate it with dispatchers, and retire unused resources.

1. The stated objective is a faster learning cycle, not maximum infrastructure consumption. A measure covering provisioning, user evaluation and retirement tests whether teams can experiment and stop paying for unnecessary capacity. Cloud services can support this agility, but a deployment label alone does not demonstrate business value. Cost, access and reliability still need their own evidence; this measure makes no automatic claim about them.

Answer: Cloud Storage for media objects organized in buckets.

1. The repository's unit of access is an entire video or image rather than a relational row or analytical aggregate. Cloud Storage organizes objects in buckets and supports upload and retrieval of unstructured content, which matches those explicit access requirements.
2. A separate catalog already handles metadata, so there is no stated need to make the media repository a relational application database. This is a choice about access model and workload, not a claim that database products can never hold file-related information or participate in a larger media architecture.

Answer: Test preservation of required case facts and measure supervisor review effort on representative handovers.

1. Frame success around what the summaries must preserve and what work they should reduce. The omitted complaint is a specific failure, so the evaluation needs task-specific criteria checking required facts alongside a measure of the supervisor's actual review effort.
2. Brevity, general benchmarks, and throughput can inform secondary tradeoffs, but none directly establishes a safe, useful handover. The Gen AI evaluation service supports tailored criteria and custom evaluation logic; the team must define and test those criteria rather than treating a general model score as evidence for this workflow.

Answer: Run the supported Windows environment on Compute Engine virtual machines.

1. Start with the stated compatibility and timing constraints. This move must preserve a vendor-supported Windows environment with administrator control over operating-system components; application redesign is excluded. Compute Engine supplies self-managed VMs, including Windows environments, so it fits that initial migration. Managed containers or a SaaS replacement could be evaluated later, but require application or workflow changes outside this scope. The firm still needs to manage its guest configuration and application security.

Answer: Grant the workload's service account the suitable read-only role on the dataset.

1. Identify the principal, the allowed action and the resource scope independently. A service account represents the workload, the read-only role supplies the needed permissions, and the dataset is the narrow resource on which that role can be granted. Broader project or organization grants add access not requested by the job. A human administrator identity also fails the explicit workload-identity requirement. Effective access should still be checked for inherited grants before accepting the configuration.

Answer: An alerts-only budget notifies; automatic restriction requires a separately configured supported control.

1. The scenario names an alerts-only budget and explicitly excludes automation and a spend-cap configuration. Its thresholds inform recipients about spending; they do not themselves stop usage or charges. Continued rendering is therefore consistent with this setup. The team can investigate separate cost-control responses and current spend-cap budgets for supported services and projects, rather than generalizing that budgets can never cap spend. Any selected restriction needs its own configuration and evaluation of operational consequences.

Answer: Run the application on IaaS virtual machines with guest operating system control.

1. The decisive constraints are the existing licensed application, a particular operating system image, and a kernel driver. These require control below an application runtime. IaaS supplies virtualized infrastructure while allowing the customer to maintain the guest software stack.
2. Virtual machines therefore fit the stated division of responsibility: the provider operates physical infrastructure, and the customer administers the guest. A managed application runtime would require checking or changing the software requirements, while retaining local servers would leave physical operations with the company.

Answer: Pub/Sub for ingestion, Dataflow for transformation, BigQuery for analytical storage, and Looker for dashboards.

1. The business change is from delayed reporting to a continuously updated analytical pipeline. In the documented service pattern, Pub/Sub receives external messages, Dataflow reads and transforms them, BigQuery provides the analytical store, and Looker presents business intelligence from that data.
2. These roles form a pipeline rather than interchangeable names for one database. The arrangement can support timely decisions, but actual end-to-end freshness still depends on configuration and operating behavior. Selecting the appropriate services does not establish a measured latency target or prove that a disruption will be detected.