Google Cloud Professional Cloud Architect Data, Analytics, and AI Solution Architecture Syllabus

Data, Analytics, and AI Solution Architecture syllabus — full chapter and topic list

Expand any chapter to see its topics and sub-topics. This is the whole examinable outline for Data, Analytics, and AI Solution Architecture in Google Cloud Professional Cloud Architect, not a summary of it.

1. 1

   Data Pipeline and Ingestion Design

   4 topics

   • Batch vs. streaming ingestion patterns
   • Pub/Sub for messaging and event streaming
   • Dataflow for stream and batch processing
     • Apache Beam programming model
   • Data Fusion and Datastream
2. 2

   Analytics and Data Warehousing

   4 topics

   • BigQuery architecture and design
     • Partitioning and clustering
     • Capacity vs. on-demand pricing
   • Dataproc for managed Hadoop and Spark
   • Looker and BI integration
   • Data governance with Dataplex
3. 3

   Database Solution Selection

   4 topics

   • Choosing the right database for the workload
     • OLTP vs. OLAP vs. NoSQL trade-offs
   • Spanner for global consistency
   • Bigtable for high-throughput workloads
   • Firestore and Memorystore use cases
4. 4

   AI/ML Solution Integration

   4 topics

   • Vertex AI platform overview
     • Training, deployment, and pipelines
   • Pre-trained APIs and generative AI services
   • Architecting for ML workloads and accelerators
   • MLOps and model lifecycle management

Data, Analytics, and AI Solution Architecture flashcards for Google Cloud Professional Cloud Architect

21 of 51 cards from the Data, Analytics, and AI Solution Architecture deck — real questions with worked answers.

1. What is the core difference between batch and streaming ingestion?

   Batch ingestion collects and processes data in large, bounded groups on a schedule (high latency, high throughput); streaming ingestion processes unbounded data continuously as events arrive (low latency, real-time).

2. On Google Cloud, which service combination is the canonical pattern for real-time streaming ingestion and processing?

   Pub/Sub (ingest/buffer events) feeding into Dataflow (stream processing), often landing in BigQuery or Bigtable.

3. What are common batch ingestion patterns on Google Cloud?

   Loading files from Cloud Storage into BigQuery (load jobs), Dataproc batch Spark/Hadoop jobs, and scheduled Dataflow batch pipelines.

4. What is Pub/Sub and what messaging model does it use?

   Pub/Sub is a fully managed, global, asynchronous messaging service using the publish/subscribe model: publishers send messages to topics, and subscribers receive them via subscriptions.

5. What is the difference between a Pub/Sub pull and push subscription?

   Pull: subscriber clients request (pull) and acknowledge messages on their own schedule. Push: Pub/Sub delivers messages via HTTP POST to a subscriber endpoint and uses the HTTP response as the ack.

6. What delivery guarantee does Pub/Sub provide by default, and what feature reduces duplicates?

   At-least-once delivery (duplicates possible). Exactly-once delivery can be enabled per subscription, and message ordering keys preserve order within a key.

7. What is the difference between Pub/Sub and Pub/Sub Lite?

   Pub/Sub is global, fully managed, and auto-scaling. Pub/Sub Lite is zonal, requires pre-provisioned capacity (throughput/storage), and is much cheaper for high-volume, predictable workloads.

8. What is a Pub/Sub dead-letter topic used for?

   It captures messages that fail to be processed after a configured maximum number of delivery attempts, preventing them from blocking the subscription and allowing later inspection.

9. What is Dataflow and what programming model does it use?

   Dataflow is a fully managed, serverless service for unified stream and batch data processing based on the open-source Apache Beam programming model.

10. What single capability makes Dataflow distinctive compared with separate batch and stream engines?

    A unified (Beam) model: the same pipeline code can run on bounded (batch) or unbounded (streaming) data, with autoscaling and dynamic work rebalancing.

11. In Dataflow streaming, what are windowing and watermarks?

    Windowing divides an unbounded stream into finite chunks (fixed, sliding, session) for aggregation; the watermark is Dataflow's estimate of event-time progress used to decide when a window's data is complete.

12. What is Cloud Data Fusion?

    A fully managed, code-free, graphical data integration (ETL/ELT) service built on open-source CDAP for visually building and managing data pipelines with prebuilt connectors and transformations.

13. What is Datastream and its primary use case?

    Datastream is a serverless change data capture (CDC) and replication service that streams changes from operational databases (e.g., Oracle, MySQL, PostgreSQL) into BigQuery, Cloud Storage, or other targets in near real time.

14. When would you choose Data Fusion over Dataflow?

    Choose Data Fusion for visual, no-code/low-code ETL built by data engineers/analysts; choose Dataflow for custom, code-based, large-scale stream/batch processing with the Beam model.

15. What is BigQuery and what type of architecture does it use?

    BigQuery is a serverless, fully managed, petabyte-scale data warehouse that separates storage and compute, using a columnar storage format and a distributed query engine (Dremel).

16. What two internal technologies underpin BigQuery's separation of compute and storage?

    Dremel (the distributed query execution engine providing 'slots' of compute) and Colossus (Google's distributed file system for columnar storage), connected by the Jupiter network.

17. What is a BigQuery slot?

    A slot is a unit of computational capacity (virtual CPU + memory) used to execute SQL queries; concurrency and performance scale with the number of available slots.

18. What are BigQuery's two main pricing models for compute?

    On-demand pricing (billed per bytes scanned by queries) and capacity/editions pricing (billed for reserved or autoscaling slots over time).

19. How do partitioning and clustering improve BigQuery performance and cost?

    Partitioning splits a table by a column (date/ingestion time/integer range) so queries prune whole partitions; clustering sorts data within partitions by chosen columns so queries scan fewer blocks—both reduce bytes scanned.

20. What is BigQuery BI Engine?

    An in-memory analysis acceleration layer that caches data for sub-second query response, speeding up dashboards in Looker, Looker Studio, and other BI tools.

21. What is Dataproc?

    A fully managed service for running open-source Apache Hadoop, Spark, Hive, Presto, and related tools on Google Cloud, with fast cluster provisioning (typically under 90 seconds).

See more Data, Analytics, and AI Solution Architecture flashcards →

Planning Data, Analytics, and AI Solution Architecture for Google Cloud Professional Cloud Architect

Data, Analytics, and AI Solution Architecture (Google Cloud Professional Cloud Architect) FAQ