AWS Certified SysOps Administrator – AssociateReliability and Business ContinuityEasy
A global e-commerce application uses Amazon DynamoDB for its product catalog. The application experiences unpredictable traffic spikes, with reads occasionally exceeding 100,000 RCU and writes up to 20,000 WCU during peak sales events. The company needs to ensure consistent low-latency performance for both reads and writes, even during these spikes, without over-provisioning capacity unnecessarily during off-peak times. Which DynamoDB capacity mode should the SysOps Administrator recommend?
- AGlobal Tables with provisioned capacity.
- BProvisioned capacity mode without auto scaling.
- CProvisioned capacity mode with auto scaling.
- DOn-demand capacity mode.
Show answer & explanationAnswer & explanation
Correct answer: D. On-demand capacity mode.
On-demand capacity mode automatically scales read and write capacity based on actual traffic patterns, making it ideal for unpredictable workloads and ensuring consistent performance without manual intervention or over-provisioning.
Why the other options are wrong
- A. Global Tables are for multi-region replication and high availability, not primarily for optimizing capacity for a single-region, unpredictable workload.
- B. Provisioned capacity without auto scaling would require constant manual adjustments or significant over-provisioning to handle unpredictable spikes.
- C. Provisioned capacity with auto scaling can react to changes but still involves setting min/max values and can have a slight delay in scaling up during sudden, sharp spikes compared to on-demand.
DynamoDB On-demand Capacity
DynamoDB On-demand capacity mode allows you to pay for read and write requests as they occur, providing automatic scaling for unpredictable workloads without requiring capacity planning.
- Pay-per-request pricing.
- Automatically scales up and down.
- Ideal for unpredictable traffic patterns.
- Can handle sudden spikes efficiently.
Memory trick: On-Demand is like a taxi, Provisioned is like a bus route.