AWS Certified Data Analytics – Specialty — Question 150

A company needs to implement a near-real-time messaging system for hotel inventory. The messages are collected from 1,000 data sources and contain hotel inventory data. The data is then processed and distributed to 20 HTTP endpoint destinations. The range of data size for messages is 2-500 KB.
The messages must be delivered to each destination in order. The performance of a single destination HTTP endpoint should not impact the performance of the delivery for other destinations.
Which solution meets these requirements with the LOWEST latency from message ingestion to delivery?

Answer options

• A. Create an Amazon Kinesis data stream, and ingest the data for each source into the stream. Create 30 AWS Lambda functions to read these messages and send the messages to each destination endpoint.
• B. Create an Amazon Kinesis data stream, and ingest the data for each source into the stream. Create a single enhanced fan-out AWS Lambda function to read these messages and send the messages to each destination endpoint. Register the function as an enhanced fan-out consumer.
• C. Create an Amazon Kinesis Data Firehose delivery stream, and ingest the data for each source into the stream. Configure Kinesis Data Firehose to deliver the data to an Amazon S3 bucket. Invoke an AWS Lambda function with an S3 event notification to read these messages and send the messages to each destination endpoint.
• D. Create an Amazon Kinesis data stream, and ingest the data for each source into the stream. Create 20 enhanced fan-out AWS Lambda functions to read these messages and send the messages to each destination endpoint. Register the 20 functions as enhanced fan-out consumers.

Correct answer: B

Explanation

Option B is the correct choice because using a single enhanced fan-out AWS Lambda function ensures that messages are processed and sent quickly to all destinations without performance degradation, while preserving the order of message delivery. In contrast, option A may introduce latency due to multiple functions, option C involves a delay with S3, and option D, while effective, uses more resources than necessary for the requirements.