AIP-C01 · Question #5
A company is developing a generative AI (GenAI) application that uses Amazon Bedrock foundation models. The application has several custom tool integrations. The application has experienced unexpected
The correct answer is C. Use Amazon CloudWatch Logs to capture model invocation logs. Create CloudWatch metric. The critical requirement is automatically adjusting thresholds as traffic patterns change. Amazon CloudWatch Anomaly Detection uses machine learning to learn the normal baseline of a metric over time and automatically adjusts its alert thresholds as patterns evolve - no manual th
Question
A company is developing a generative AI (GenAI) application that uses Amazon Bedrock foundation models. The application has several custom tool integrations. The application has experienced unexpected token consumption surges despite consistent user traffic. The company needs a solution that uses Amazon Bedrock model invocation logging to monitor InputTokenCount and OutputTokenCount metrics. The solution must detect unusual patterns in tool usage and identify which specific tool integrations cause abnormal token consumption. The solution must also automatically adjust thresholds as traffic patterns change. Which solution will meet these requirements?
Options
- AUse Amazon CloudWatch Logs to capture model invocation logs. Create CloudWatch
- BStore model invocation logs in Amazon S3. Use AWS Glue and Amazon Athena to analyze token
- CUse Amazon CloudWatch Logs to capture model invocation logs. Create CloudWatch metric
- DStore model invocation logs in an Amazon S3 bucket. Use AWS Lambda to process logs in real
How the community answered
(23 responses)- A4% (1)
- B13% (3)
- C78% (18)
- D4% (1)
Explanation
The critical requirement is automatically adjusting thresholds as traffic patterns change. Amazon CloudWatch Anomaly Detection uses machine learning to learn the normal baseline of a metric over time and automatically adjusts its alert thresholds as patterns evolve - no manual threshold tuning is required. By creating CloudWatch metric filters on the invocation logs to extract InputTokenCount and OutputTokenCount per tool integration, and then applying Anomaly Detection alarms, the solution detects unusual per-tool token spikes and self-adjusts as traffic grows or changes. Option A uses static CloudWatch alarms with fixed thresholds, which cannot auto-adjust. Option B (S3 + Glue + Athena) is a batch analytics approach with high latency, unsuitable for real-time detection. Option D (S3 + Lambda) adds operational overhead and lacks the built-in auto-threshold adjustment of Anomaly Detection.
Topics
Community Discussion
No community discussion yet for this question.