MLS-C01 · Question #241
A company is building a pipeline that periodically retrains its machine learning (ML) models by using new streaming data from devices. The company's data engineering team wants to build a data…
The correct answer is A. Configure the devices to send streaming data to an Amazon Kinesis data stream. Configure an. Option A configures devices to send data to an Amazon Kinesis Data Stream, then uses an AWS Lambda function (triggered by the stream) to apply basic transformations before writing results to S3. Kinesis Data Streams provides high throughput, durability (24-hour default…
Question
A company is building a pipeline that periodically retrains its machine learning (ML) models by using new streaming data from devices. The company's data engineering team wants to build a data ingestion system that has high throughput, durable storage, and scalability. The company can tolerate up to 5 minutes of latency for data ingestion. The company needs a solution that can apply basic data transformation during the ingestion process. Which solution will meet these requirements with the MOST operational efficiency?
Options
- AConfigure the devices to send streaming data to an Amazon Kinesis data stream. Configure an
- BConfigure the devices to send streaming data to an Amazon S3 bucket. Configure an AWS
- CConfigure the devices to send streaming data to an Amazon S3 bucket. Configure an AWS Glue
- DConfigure the devices to send streaming data to an Amazon Kinesis Data Firehose delivery
How the community answered
(48 responses)- A69% (33)
- B8% (4)
- C4% (2)
- D19% (9)
Explanation
Option A configures devices to send data to an Amazon Kinesis Data Stream, then uses an AWS Lambda function (triggered by the stream) to apply basic transformations before writing results to S3. Kinesis Data Streams provides high throughput, durability (24-hour default retention, up to 7 days), and elasticity via shard scaling. Lambda handles lightweight transformation serverlessly with no infrastructure management. The 5-minute latency tolerance easily accommodates this near-real-time pipeline. Options B and C use S3 as the entry point, which adds latency and lacks true streaming semantics. Option D (Kinesis Data Firehose) is also a good pattern, but Firehose has a minimum buffer time of 60 seconds and the answer key indicates A (Kinesis Data Streams + Lambda) as the most operationally efficient approach with the right combination of streaming ingestion and transformation.
Topics
Community Discussion
No community discussion yet for this question.