Cloud Digital Leader Why cloud technology is transforming business Practice Question
A multinational manufacturing company operates thousands of IoT sensors on factory equipment. These sensors generate over 50 TB of telemetry data daily. The company wants to implement predictive maintenance to reduce unplanned downtime. Their current on-premises infrastructure is maxed out, and they have a small IT team with limited data engineering expertise. They are evaluating cloud vs. on-premises options. The data is highly sensitive and must be encrypted at rest and in transit. Additionally, they need to run machine learning models near real-time and store historical data for trend analysis. The CTO is concerned about vendor lock-in, data sovereignty, and the ability to scale globally as they open new factories. Which course of action best addresses these requirements using Google Cloud?
⚠ Common exam trap
Google Cloud often tests the misconception that batch processing (e.g., BigQuery on stored data) can substitute for a streaming pipeline in near-real-time scenarios, leading candidates to overlook the need for continuous ingestion and processing with services like Pub/Sub and Dataflow.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Use Cloud Pub/Sub to ingest sensor data, Cloud Dataflow for stream processing, BigQuery for historical storage and analysis, and Vertex AI for predictive models, all in the desired region.
Use Cloud Pub/Sub to ingest streaming sensor data, Cloud Dataflow for processing, BigQuery for historical storage and analysis, and Vertex AI for predictive models. This provides near-real-time ML, encryption at rest and in transit, regional data sovereignty, and global scalability.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Deploy Kubernetes on-premises to orchestrate microservices, and use Cloud Storage for backup only.
Why it's wrong here
Running Kubernetes on-premises still requires capacity planning, hardware procurement, and ongoing cluster maintenance, which directly conflicts with the company's lack of staff and need for elasticity. Using Cloud Storage only for backups leaves the core ingestion and analytics requirements unaddressed, and an on-prem cluster cannot automatically scale to handle millions of concurrent sensor messages. This option also fails to provide a managed path to real-time ML predictions, as it lacks the integrated streaming and AI services needed.
- ✗
Build a new on-premises data center with dedicated GPU servers for ML training, and hire additional data engineers.
Why it's wrong here
Building a new on-premises data center with dedicated GPU servers involves significant capital expenditure and a long deployment timeline, and hiring more data engineers increases operational burden rather than eliminating it. GPU clusters are also notoriously underutilized during idle periods, meaning the company would pay for fixed capacity instead of scaling with actual IoT workloads. Unlike a managed cloud service, this approach cannot quickly spin up resources in the desired region to meet compliance and residency requirements.
- ✓
Use Cloud Pub/Sub to ingest sensor data, Cloud Dataflow for stream processing, BigQuery for historical storage and analysis, and Vertex AI for predictive models, all in the desired region.
Why this is correct
This fully managed pipeline uses Cloud IoT Core to securely connect and ingest sensor telemetry, then Pub/Sub provides durable, low-latency message buffering with decoupled consumers. Dataflow performs stateful, exactly-once stream processing and windowing for anomaly detection, while AI Platform trains and serves predictive models; deploying all services in the desired region satisfies data residency and compliance. Because each component is serverless and auto-scaling, the solution handles massive throughput without infrastructure management, directly addressing the company's constraints.
- ✗
Store all sensor data in Cloud Storage and run ad-hoc queries using BigQuery without any streaming pipeline.
Why it's wrong here
Simply dumping raw sensor data into Cloud Storage and running ad-hoc BigQuery queries is a batch architecture, so it cannot deliver the real-time insights or streaming anomaly detection the manufacturer needs. While BigQuery supports continuous data ingestion via the Storage Write API, it is not designed for low-latency ML inference, and querying arbitrary raw blobs would require repeated full scans, increasing cost and latency. Without a streaming pipeline and a model-serving layer, predictive maintenance and immediate response to equipment failures are impossible.
Go deeper
Related to this question
Learn chapter
Open Source and Open Standards at Google Cloud
Key term
Scalability
Scalability is the ability of a system, network, or process to handle a growing amount of work by adding resources, either by making the existing resources more powerful (vertical scaling) or by adding more resources (horizontal scaling).
Key term
Pub/Sub
Pub/Sub is a messaging pattern where publishers send messages without knowing who receives them, and subscribers receive only the messages they care about.
About these practice questions
This GCDL question is part of Courseiva's 848-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This GCDL practice question is part of Courseiva's free Google Cloud certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the GCDL exam.