Weighing ACID guarantees and availability against each other
ACID compliance and availability are key criteria for choosing a Google Cloud data store, and the exam expects you to match workload requirements (transactional integrity vs. massive scale/availability) to the right product. Understanding which services offer full ACID guarantees versus eventual consistency or limited transactional scope is essential for designing correct, resilient data pipelines.
1 · Learn the must-know
- Cloud Spanner offers strong external consistency and full ACID transactions (including multi-row, multi-table) with horizontal scalability and high availability (up to 99.999% with multi-region configs), making it the choice when you need both ACID and global scale.
- Cloud SQL provides full ACID compliance via standard relational engines (MySQL, PostgreSQL, SQL Server) but is limited to regional availability and vertical scaling, so it's not suited for workloads needing horizontal write scale or multi-region durability.
- BigQuery supports ACID transactions at the row level within a single table via DML statements (INSERT/UPDATE/DELETE/MERGE), but it is not designed for high-frequency transactional workloads, it's optimized for analytical (OLAP) queries, not OLTP.
- Bigtable is not ACID-compliant across rows; it only guarantees atomicity for single-row mutations, making it suitable for high-throughput NoSQL use cases (time series, IoT) rather than multi-row transactional integrity.
- Firestore provides ACID transactions across multiple documents/collections and offers strong consistency plus multi-region availability, making it a good fit for mobile/web app backends needing transactional correctness at scale.
- A common exam gotcha: 'high availability' and 'ACID compliance' are independent axes, a system can be highly available but eventually consistent (e.g., Bigtable single-cluster) or strongly consistent but with lower availability guarantees than a distributed system like Spanner, so always evaluate both dimensions separately when architecting a solution.
2 · Check your understanding
A retail company is designing an order-processing system. Requirements include ACID transactions that span multiple related tables, the ability to scale write throughput beyond a single machine, and strongly consistent low-latency reads for customers in both North America and Europe. Which service should the data engineer choose?
What you have tried across GCP PDE's objectives, not a readiness score.
Designing data processing systems~22% of the exam0 of 17 tried
Ingesting and processing the data~25% of the exam0 of 11 tried
Storing the data~20% of the exam0 of 13 tried
Preparing and using data for analysis~15% of the exam0 of 11 tried
Maintaining and automating data workloads~18% of the exam0 of 15 tried
3 · Keep going
Ready for more? Take a weighted mock or try free practice questions.