GCDL

Free GCDL practice questions

10 questions from our GCDL practice tests, each with the correct answer and a written explanation. Read them with the answers shown, or switch to quiz mode and answer first.

Back to the GCDL practice exams

Question 1 of 10, Digital Transformation with Google Cloud

A government agency must keep classified workloads on infrastructure dedicated solely to it, but wants cloud-like self-service provisioning for its own teams. Which architecture BEST meets this need, and what is its main trade-off?

  1. AA private cloud, which keeps dedicated control but requires the agency to fund and operate the capacity Correct answer
  2. BA public cloud, which removes dedicated control but provides near-unlimited elastic capacity on demand
  3. CA SaaS application suite, which outsources all infrastructure and gives the agency no provisioning role
  4. DA multicloud setup, which spreads the classified data across several providers to avoid any single point of failure

Answer and explanation: A private cloud is dedicated to one organization and can provide self-service provisioning, but the organization bears the cost and effort of owning and operating the capacity. A public cloud does not meet the requirement for infrastructure dedicated solely to the agency. Spreading classified data across multiple providers conflicts with the dedicated requirement. SaaS delivers finished applications and does not offer infrastructure provisioning to the agency's teams.

Question 2 of 10, Digital Transformation with Google Cloud

What is the main difference between open source software and an open standard?

  1. AOpen source is a published specification; an open standard is software whose code anyone may modify and redistribute
  2. BOpen source makes source code freely available; an open standard is a publicly available specification anyone can implement Correct answer
  3. COpen source software can be used only on-premises, while open standards are available only inside public clouds
  4. DOpen source always requires a commercial license fee, while an open standard is maintained and owned by a single vendor

Answer and explanation: Open source refers to software whose source code is publicly available to use, modify, and share, while an open standard is a publicly documented specification that many vendors can implement. The second option reverses the two definitions. Neither concept is tied to on-premises or cloud deployment. Open source does not require license fees, and open standards are typically governed by communities or standards bodies rather than a single vendor.

Question 3 of 10, Exploring Data Transformation with Google Cloud

A global game company needs a relational database that offers strong consistency and horizontal scaling for player transactions across several continents with very high availability. Which product is the BEST fit?

  1. ACloud SQL
  2. BBigtable
  3. CSpanner Correct answer
  4. DMemorystore

Answer and explanation: Spanner is a fully managed relational database that scales horizontally and provides strongly consistent transactions across regions and continents with very high availability. Cloud SQL is relational but is designed for regional workloads and scales mainly vertically. Bigtable scales globally but is a NoSQL wide-column store without relational transactions across rows and tables. Memorystore is an in-memory cache, not a system of record for transactions.

Question 4 of 10, Exploring Data Transformation with Google Cloud

A company wants to analyze data that lives in Amazon S3 and Azure Blob Storage alongside data in Google Cloud, using one SQL interface without first copying everything into Google Cloud. Which offering supports this?

  1. AAlloyDB Omni
  2. BCloud SQL federation
  3. CStorage Transfer Service
  4. DBigQuery Omni Correct answer

Answer and explanation: BigQuery Omni lets organizations run BigQuery analytics on data stored in other clouds, such as AWS and Azure, from a single interface without moving the data first. AlloyDB Omni is a transactional PostgreSQL-compatible database for running outside Google Cloud, not a cross-cloud analytics engine. Cloud SQL federation is not designed for querying object storage on other clouds. Storage Transfer Service copies data, which the company wants to avoid.

Question 5 of 10, Innovating with Google Cloud Artificial Intelligence

A company wants one platform where its developers can access Gemini and other foundation models, ground them in company data, and build, deploy, and govern AI agents. Which Google Cloud offering is designed for this?

  1. AApigee API Management with a developer portal
  2. BCloud Run functions
  3. CLooker
  4. DGemini Enterprise Agent Platform Correct answer

Answer and explanation: Gemini Enterprise Agent Platform is Google Cloud's all-in-one AI developer platform, providing foundation models such as Gemini, tools to ground and tune them, and capabilities to build, deploy, and govern agents. Looker is a business intelligence platform. Cloud Run functions run event-driven code but are not an AI development platform. Apigee manages and publishes APIs but does not provide models or agent tooling.

Question 6 of 10, Innovating with Google Cloud Artificial Intelligence

A data analyst team is fluent in SQL and stores all customer data in BigQuery. They want to predict customer churn without exporting data or learning a new programming language. Which option meets these requirements with the LEAST effort?

  1. ASend each customer record to the Gemini API and ask it whether the customer will churn
  2. BExport the tables to Cloud Storage and train a custom TensorFlow model on GPUs
  3. CCopy the data into Cloud SQL and write stored procedures that score each customer
  4. DUse BigQuery ML to create and run the model with SQL statements inside BigQuery Correct answer

Answer and explanation: BigQuery ML lets analysts create, train, evaluate, and run ML models with standard SQL directly where the data lives, so no export or new language is needed. Exporting data and writing TensorFlow code requires ML engineering skills and data movement. Cloud SQL stored procedures are not an ML training tool and would require moving the data. Prompting a general model per record is not a trained churn model, adds cost, and does not learn from the company's labeled history.

Question 7 of 10, Modernize Infrastructure and Applications with Google Cloud

An enterprise runs workloads on Google Cloud, another public cloud, and on-premises. It wants a single platform to manage Kubernetes clusters consistently across all three, with central policy and configuration management. Which offering meets this requirement?

  1. AGoogle Cloud VMware Engine
  2. BCloud Run
  3. CCompute Engine sole-tenant nodes
  4. DGKE Enterprise Correct answer

Answer and explanation: GKE Enterprise lets organizations manage Kubernetes clusters across Google Cloud, other clouds, and on-premises with consistent policy, configuration, and service management. Cloud Run is a serverless container service rather than a multi-environment cluster management platform. VMware Engine runs VMware workloads in Google Cloud and does not manage clusters in other clouds. Sole-tenant nodes provide dedicated hardware within Google Cloud only.

Question 8 of 10, Modernize Infrastructure and Applications with Google Cloud

A company wants to run many containerized microservices with fine control over networking and scheduling, but does not want to manage the cluster's nodes. Which option fits BEST?

  1. ACompute Engine VMs with Docker
  2. BGKE in Standard mode
  3. CCloud Run functions
  4. DGKE in Autopilot mode Correct answer

Answer and explanation: GKE Autopilot provides a full Kubernetes API for orchestrating many microservices while Google manages the nodes, scaling, and security configuration. GKE Standard gives the same Kubernetes control but leaves node management to the customer. Running Docker directly on VMs requires managing both the VMs and the orchestration. Cloud Run functions run single-purpose functions and do not offer Kubernetes-level networking and scheduling control.

Question 9 of 10, Trust and Security with Google Cloud

Which practice grants users only the permissions they need to perform their job and nothing more?

  1. ASecurity by default
  2. BDefense in depth
  3. CSeparation of duties
  4. DLeast privilege Correct answer

Answer and explanation: The principle of least privilege limits each identity to the minimum permissions required for its tasks, reducing damage if an account is compromised. Defense in depth layers multiple security controls rather than limiting individual permissions. Security by default means products ship with protective settings already enabled. Separation of duties splits critical tasks across people so no single person controls an entire process.

Question 10 of 10, Scaling with Google Cloud Operations

A site reliability team defines that 99.9% of checkout requests must succeed over a rolling 30-day window. What is this target called?

  1. AService level indicator (SLI)
  2. BService level objective (SLO) Correct answer
  3. CService level agreement (SLA)
  4. DError budget burn-rate alert policy

Answer and explanation: An SLO is an internal reliability target for a measured indicator over a time window, such as 99.9% successful requests over 30 days. The SLI is the measurement itself, here the proportion of successful checkout requests. An SLA is a contract with customers that includes consequences such as credits if it is missed. An error budget burn-rate alert notifies the team when the allowed unreliability is being consumed too quickly; it is not the target itself.