MidAWS

When would you choose DynamoDB over a relational database, and how does partition key design affect performance?

What they are really testing: Whether you understand the access-pattern-first model and the hot-partition trap, the difference between someone who has run DynamoDB at scale and someone who has only read about it.

A real interview question

When would you choose DynamoDB over a relational database, and how does partition key design affect performance?

What most people say

drag me

DynamoDB is a NoSQL database that scales automatically, so use it when you need a database that handles a lot of traffic.

It is a vague "it scales" pitch. It misses the access-pattern-first modeling and the partition-key/hot-partition reality, which is where DynamoDB projects succeed or fail.

The follow-ups they ask next

  • Your DynamoDB table throttles under load even though provisioned capacity looks adequate. What is the likely cause?

    A hot partition: a skewed partition key concentrates traffic on one partition whose share of capacity is exceeded. Fix by choosing a higher-cardinality key or adding a sharding suffix to spread writes.

  • What kind of workload should stay on a relational database instead?

    Ad-hoc querying, complex joins, multi-row transactions, and reporting/analytics. DynamoDB is optimized for predefined access patterns, not flexible querying.

What the interviewer is listening for

  • Access-pattern-first modeling / single-table
  • Explains partition key distribution + hot partitions
  • Knows when relational is the better choice

What sinks the answer

  • Just "NoSQL that scales"
  • Unaware of hot partitions
  • Would use it for ad-hoc queries/joins

If you genuinely do not know

Say this instead of freezing. Reasoning out loud from what you do know beats silence every single time, and a good interviewer is listening for exactly that.

Pick DynamoDB for [known high-volume access patterns needing single-digit-ms latency at scale, not ad-hoc queries/joins]. You [model around access patterns up front, often single-table with composite keys + GSIs]. The [partition key drives distribution]: [high-cardinality spreads load, a skewed key creates a hot partition that throttles]. Use [on-demand vs provisioned, eventual vs strong reads].

Keep going with aws

All 336 cloud engineer questions

Knowing the answer is not the same as recalling it under pressure

Sign in to send the questions you fumble to spaced recall, so they come back right before you would forget them, and learn the concepts behind them with hands-on labs.

Start free