AWS Certified SysOps Administrator – AssociateCost and Performance OptimizationMedium
A company is using Amazon DynamoDB for its core application's data storage. The application experiences predictable spikes in access patterns every hour on the hour, requiring significantly higher write capacity for a short duration. The current provisioned capacity is set to handle average traffic, leading to throttling errors during these spikes. The SysOps Administrator needs to ensure that DynamoDB can handle these predictable, hourly bursts of write traffic without over-provisioning capacity for the entire hour. Which DynamoDB feature should be configured?
- AReserved Capacity.
- BAuto Scaling with target tracking.
- COn-Demand capacity mode.
- DGlobal Tables for multi-region access.
Show answer & explanationAnswer & explanation
Correct answer: B. Auto Scaling with target tracking.
DynamoDB Auto Scaling with target tracking allows you to define a target utilization level for your table's read and write capacity. When actual utilization deviates from this target, Auto Scaling automatically adjusts the provisioned capacity up or down. This is ideal for predictable spikes, as it reacts quickly to increase capacity before throttling occurs and scales back down to optimize costs.
Why the other options are wrong
- A. Reserved Capacity is a pricing option for long-term commitment, not a feature for dynamic capacity scaling to handle hourly spikes.
- C. On-Demand capacity mode handles unpredictable workloads well but can be more expensive than provisioned capacity with Auto Scaling for predictable patterns, as it charges per request.
- D. Global Tables are for multi-region, active-active replication and do not address capacity scaling for hourly spikes within a single region.
DynamoDB Auto Scaling
A feature that dynamically adjusts a DynamoDB table's provisioned read and write capacity to maintain a specified target utilization level.
- Uses AWS Auto Scaling for DynamoDB.
- Responds to actual traffic patterns.
- Optimizes performance and costs for varying workloads.
Memory trick: DynamoDB capacity: pay per use, scale automatically, or commit for less.