AWS Certified Data Engineer Associate DEA-C01 Practice Question

An ecommerce company needs to ingest millions of clickstream events per minute with sub-second latency. Multiple independent analytics applications must process the same events concurrently and occasionally replay data from the past 72 hours for backfilling. The team wants a fully managed, cost-effective solution that minimizes capacity planning. Which approach best meets these requirements?

Use Amazon Kinesis Data Firehose to batch events to Amazon S3 and trigger AWS Lambda functions for each analytics application.
Publish events to an Amazon SQS FIFO queue and have each analytics application poll the queue independently.
Write events directly to an Amazon DynamoDB table and use DynamoDB Streams to feed the analytics applications.
Send events to an Amazon Kinesis Data Streams stream configured for on-demand capacity with a 72-hour retention period; let each analytics application consume the stream through enhanced fan-out.

Report Issue

Answer Description

Amazon Kinesis Data Streams is purpose-built for high-throughput, low-latency event ingestion. With on-demand capacity mode, the service scales automatically and charges only for actual usage, removing the need to provision shards. Streams can retain data for up to 7 days (configurable) and allow multiple consumers to read the same records independently; using enhanced fan-out each consumer receives its own 2 MB/s throughput with sub-second lag, and any consumer can replay data within the retention window. SQS FIFO guarantees ordering but messages are removed after a single successful read and its throughput limits make it unsuitable for millions-per-minute traffic. Kinesis Data Firehose buffers data and delivers it to destinations but does not provide native multi-consumer access or replay once data is delivered. DynamoDB with Streams offers only 24-hour retention, lower write throughput for this scale, and additional operational overhead for partition management.

Ask Bash

Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.