An application writes clickstream data to Amazon Kinesis. The stream experiences periodic spikes and PutRecords calls sometimes fail with the shown response. Which techniques can help mitigate this error? (Choose two.)

The error indicates throttling, meaning the Kinesis stream's write capacity has been exceeded. Implementing retries with exponential backoff is a standard and effective technique for handling transient errors like throttling. This allows the application to automatically reattempt failed requests after increasing delays, giving the stream time to recover. Reducing the frequency and/or size of requests directly addresses the root cause of throttling by lowering the write load on the stream. Using PutRecord instead of PutRecords is less efficient as it sends one record at a time, potentially increasing the number of API calls and exacerbating throttling. Replacing Kinesis with Amazon SNS is not a direct solution for Kinesis throttling; SNS is a messaging service, not a streaming data platform for clickstream data. Reducing KCL consumers affects reading from the stream, not writing to it, and is therefore irrelevant to this throttling issue.