What questions would you ask before choosing a database technology for a new system?
Quick Answer
Ask about the data's shape and relationships, expected scale (both current and projected growth) for reads/writes/storage, consistency requirements (what's the real cost of stale or lost data for this specific data), the team's existing operational expertise, budget/licensing constraints, and any compliance/regulatory requirements. The goal is gathering enough concrete information to make a reasoned choice, rather than defaulting to whatever's most familiar or currently trendy.
Detailed Answer
This question tests whether a candidate gathers information deliberately before making an architectural decision, rather than pattern-matching to a favorite tool. A strong answer organizes the questions into clear categories.
Questions about the data itself
- What does the data actually look like — fixed and well-structured, or highly variable/nested per record?
- What are the real relationships between entities — how relational is this data, genuinely (not just "everything is technically related to everything")?
- How will this data be queried — simple key lookups, complex ad-hoc joins/aggregation, full-text search, graph traversal, geospatial queries?
Questions about scale and growth
- What's the expected read/write volume today, and realistically in 1-2 years — not a wildly speculative "what if we go viral" number, but a grounded estimate?
- Is the workload read-heavy, write-heavy, or roughly balanced?
- How much total data storage is expected, and how fast does that grow?
Questions about consistency and correctness requirements
- What's the actual cost of a stale read, or (worse) a lost write, for this specific data? (A financial balance and a "like" counter have very different answers.)
- Are there multi-entity invariants that must be enforced atomically (e.g., "total allocated seats never exceeds capacity")?
Questions about team and organizational context
- What does the team already have deep operational experience running and troubleshooting in production? (Adopting an unfamiliar technology has a real, often underestimated, ongoing cost.)
- What's the existing infrastructure/cloud-provider ecosystem, and what does it make easiest to operate reliably?
Questions about constraints
- What's the budget for licensing/managed-service costs?
- Are there compliance requirements (HIPAA, PCI-DSS, data residency laws) that constrain where/how data can be stored, encrypted, or audited?
- Are there existing systems this needs to integrate with that constrain the choice (e.g., an existing data warehouse pipeline that expects a specific format)?
Why asking questions (rather than jumping to an answer) is itself the right signal
A candidate who immediately says "use Postgres" or "use MongoDB" without first asking any of the above is skipping the actual analysis a real architectural decision requires — the "right" database is entirely dependent on answers to these questions, and a senior engineer's value in this conversation is in knowing which questions actually change the recommendation, not in having a single favorite tool for every situation.