DEA-C01 Practice Question 45: DynamoDB Capacity Mode Choice

On-demand is "a serverless throughput option that simplifies database management and automatically scales to support customers' most demanding applications." The documentation states that it "enables you to create a table without worrying about capacity planning, monitoring usage, and configuring scaling policies", charges per request so "you only pay for what you use", and is "the default and recommended throughput option for most DynamoDB workloads", scaling "to support workloads that can start small and scale to millions of requests per second." That matches an unforecastable launch exactly, and the mode can be changed later once real traffic data exists.

Option B pays for the worst case every hour of every day. In provisioned mode "you'll be charged based on the hourly read and write capacity you have provisioned, not how much of that provisioned capacity you actually consumed", so a table sized for a spike that may never come is pure waste, and if the estimate is too low, customers are throttled anyway.

Option C is the most tempting wrong answer, because auto scaling is a real mitigation. But it reacts to CloudWatch alarms and takes minutes to adjust, so a sudden marketing-driven spike throttles while scaling catches up. It also still requires someone to set target utilization and bounds, the capacity modelling the team wanted to skip.

Option D optimizes the price of capacity the team cannot yet size. Reserved capacity is a one-year commitment suited to steady, well-understood baseline traffic; committing before launch risks paying for unused capacity for a year.

Reference: DynamoDB throughput capacity