Objective: start with the data operation
Using the location and responsibility distinctions from earlier lessons, explain what a workload must do with its data before naming a product. Retaining image files, updating an order, and comparing annual sales all involve data, but they ask for different operations. State the information stored, how users read or change it, and which requirement would make a choice unsuitable. Data volume matters, yet volume by itself does not determine whether the workload needs object storage, a transactional database, or an analytics platform.
Objects and operational records
Cloud Storage stores objects in buckets. It is a candidate for retained files such as product photographs or exported documents, with permissions controlling who can access them. A relational application database organizes structured records that an application queries and updates. Cloud SQL supplies managed MySQL, PostgreSQL, and SQL Server databases. Its managed operation does not design the application’s schema or decide which records a user may see. Do not confuse a file name pointing to an object with a complete relational order-processing system.
Scale and analytics are separate requirements
Spanner is relevant when the application needs strongly consistent transactions at scale. The decision needs explicit transaction, consistency, and distribution requirements; “we might grow” is insufficient to establish that need. BigQuery is a managed analytics platform with SQL and Python interfaces and a serverless architecture. It is a candidate when analysts must explore and aggregate data. Avoid the false shortcut that analytical systems never support transactions: instead compare the workload purpose, access pattern, and service capabilities. Product categories guide investigation without replacing service-specific checks.
Worked example
Fictional case: a furniture seller retains product photographs, records purchases in an existing PostgreSQL application, and asks analysts to compare three years of sales. Separate the three operations. Evaluate Cloud Storage for the photo objects. Investigate Cloud SQL for the managed PostgreSQL workload, checking the application’s actual compatibility needs. Investigate BigQuery for historical analytical queries. These components can cooperate: a transaction record can reference a photograph, and selected sales data can feed analysis. The result is a division of roles, not a demand to store every item in three services.
Challenge a tempting replacement
Now the seller requests transactions across a widely distributed, rapidly growing application with an explicit strong-consistency requirement. That new constraint creates a reason to evaluate Spanner. It does not show that photograph storage or historical analysis should automatically move there. Conversely, if the actual problem is only slow reporting on the operational database, investigate separating analysis before redesigning all transactions. Each change should address the observed operation. A product name is not a substitute for measuring the current workload and testing compatibility.
Transfer check
A legal publishing team needs to retain finished PDF files and let approved staff download them. No record-level transaction or analytical-query requirement is stated. Which starting category fits? Check: object storage is the supported starting point; access policies and retention requirements still need design. Explain why “documents contain text” does not establish a need for a relational database. Use this unit’s cards to retrieve the four service roles, then continue to the analytics-pipeline lesson to learn what happens between a data source and an analytical result.
Try it, then explain it.
Complete a short set without looking at the example. After each answer, explain which evidence or rule supports your decision and why the alternatives do not fit.
Sources & authorship
Original OpenExam explanations and examples. References support the concepts; they are not the source of copied questions. Independent educator review is pending.
BigQuery overview | Google Cloud DocumentationCloud SQL overview | Google Cloud DocumentationSpanner: Always-on, virtually unlimited scale database | Google CloudCloud Storage overview | Google Cloud Documentation