Question #749
A social media platform uses Amazon Kinesis Data Streams to process real-time user activity data. Multiple consumer applications are reading from the stream for analytics and monitoring. However, some consumers are experiencing throttling and intermittently receiving a ReadProvisionedThroughputExceeded error. What should a solutions architect recommend to resolve this issue? (Choose three.)
Reshard the stream to increase the number of shards.
Use consumers with the enhanced fan-out feature.
Implement an error retry and exponential backoff mechanism in the consumer logic.
Reshard the stream to decrease the number of shards.
Use the Kinesis Producer Library (KPL) and adjust the batching settings.
Configure the stream to use on-demand capacity mode.
Explanation
The ReadProvisionedThroughputExceeded error occurs when consumers exceed the read limits of Kinesis Data Streams in provisioned mode.
- A. Resharding increases the number of shards, raising read capacity (each shard allows 5 reads/sec and 2 MB/s).
- B. Enhanced fan-out gives consumers dedicated throughput (2 MB/s per consumer), avoiding contention.
- C. Retry with backoff handles throttling by retrying failed reads with delays.
Other options:
- D. Decreasing shards worsens throughput.
- E. KPL affects producers, not consumers.
- F. On-demand mode auto-scales shards, but the answer assumes staying in provisioned mode. While F is valid, the exam's answer focuses on ABC.
Answer
The correct answer is: ABC