Scalable Streaming Analytics for Clickstream Data
Short answer
C Use Pub/Sub for ingestion, Dataflow with Apache Beam for processing, and BigQuery for storage and analytics.
Question
Key Takeaways
The key trap is confusing batch processing tools (like Dataproc/Spark) with streaming tools; only Dataflow with Apache Beam provides native, managed support for stateful windowing and late-arriving data in a continuous stream.
This page explains why Pub/Sub, Dataflow, and BigQuery form the optimal architecture for low-latency, scalable clickstream processing with sessionization capabilities on Google Cloud Platform.
Common mistake: Candidates often select Dataproc (Option D) because Spark is powerful for big data, but they overlook the explicit requirement for 'low latency' and 'continuous' processing, where serverless Dataflow is the superior managed service compared to cluster-managed Spark.
Expert Analysis
The question requires a solution for continuous, low-latency processing of high-volume clickstream data that includes complex transformations like sessionization and aggregation. Crucially, the system must be scalable to millions of events per second and resilient to late-arriving data.
Option C is the correct choice because it leverages Google Cloud's managed serverless streaming pipeline: Pub/Sub provides infinite scalability for ingestion; Dataflow (using Apache Beam) natively supports stateful operations required for sessionization and windowing with built-in handling for late data; and BigQuery offers scalable storage and SQL-based analytics for the processed results.
Option A fails because Firebase Realtime Database lacks the throughput for millions of events per second, and Cloud Run functions are not designed for stateful, long-running stream processing tasks like sessionization. Option B is incorrect as Cloud Data Fusion is primarily an orchestration tool rather than a direct processing engine, and Cloud SQL is a relational database unsuited for high-throughput streaming analytics or large-scale aggregation. Option D utilizes Dataproc with Spark, which is optimized for batch processing rather than low-latency continuous streams, making it unsuitable for the "low latency" and "continuous" requirements specified.
Exam Strategy
When evaluating cloud architecture options, strictly match the processing pattern to the service capabilities. Keywords like 'continuous', 'low latency', and 'late-arriving data' point exclusively to managed streaming services (e.g., Dataflow). If the question mentioned 'historical analysis' or 'daily reports', batch solutions (e.g., Dataproc/Data Proc) might be appropriate, but never for real-time stream requirements.
Frequently Asked Questions
Why can't I use Cloud Run for this task?
What makes Dataflow better than Dataproc for this scenario?
Community Votes
100% of anonymous learners picked answer C. Votes are pick records left by other test-takers — they are not the verified answer.
Discussion 2 contributions
Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.
From public community threads 2
-
AwesomeGCP 👍 1 Answered CThe correct answer is C. Use Pub/Sub for ingestion, Dataflow with Apache Beam for processing, and BigQuery for storage and analytics.
-
hola_prep 👍 1 Answered CC. Use Pub/Sub for ingestion, Dataflow with Apache Beam for processing, and BigQuery for storage and analytics.
Reader comments & corrections
No reader corrections yet — spotted an error or have a note? Share it below.