Amazon DynamoDB Unlocked (2025): Mastering Core Components ...

3. **Pitfall: Misunderstanding Consistency Models.** - **Problem:** Not realizing that reads from GSIs and default table reads are eventually consistent. This can lead to stale data if an immediate read follows a write. - **Best Practice:** Understand the trade-offs. For reads that require absolute up-to-date data, use strongly consistent reads on the base table (doubles RCU cost). For GSIs, design your application to tolerate eventual consistency or implement read-after-write patterns with retries/delays if strong consistency is critical on GSI data. 4. **Pitfall: Over/Under-provisioning Capacity (Provisioned Mode).** - **Problem:** Setting RCUs/WCUs too low leads to throttling and errors. Setting them too high wastes money. - **Best Practice:** Monitor `Consumed*CapacityUnits` and `ThrottledRequests` metrics in CloudWatch. Implement DynamoDB Auto Scaling to adjust provisioned capacity automatically based on demand. Consider On-Demand mode if workloads are highly unpredictable. … - **Problem:** Applying relational database design principles (multiple normalized tables) directly to DynamoDB can lead to many tables and complex application-side joins, negating NoSQL benefits. - **Best Practice:** For simple use cases, separate tables are fine. For complex, related data, explore **Single Table Design (STD)**. This involves storing different but related entity types in a single table, using generic primary key attributes (e.g., `PK`, `SK`) and GSIs to model one-to-many and many-to-many relationships. This is an advanced topic but incredibly powerful.