AWS Certified AI Practitioner

Explanation

Why Option C is correct:

1. OpenSearch Serverless optimization: When using Amazon Bedrock Knowledge Bases, the vector database (OpenSearch Serverless) is a critical component for storing and retrieving embeddings. Properly sizing capacity units and configuring index refresh intervals directly impacts query performance.

2. Capacity units: OpenSearch Serverless uses capacity units (CUs) to allocate resources. If the capacity is undersized for the workload, query performance will suffer. Properly sizing these units ensures adequate compute and memory resources for efficient retrieval.

3. Index refresh intervals: The refresh interval determines how frequently new data becomes searchable. A longer refresh interval can improve query performance by reducing the frequency of index updates, though it may increase data latency.

Why other options are incorrect:

• Option A (Reduce model temperature): Temperature controls the randomness/creativity of text generation, not retrieval speed. It doesn't affect embedding retrieval performance.

• Option B (Use a smaller embedding dimension size): While smaller embeddings might theoretically be faster to process, this isn't the most effective approach. Embedding dimensions are typically fixed by the model, and changing them would require retraining or using a different model, which isn't practical for improving existing system performance.

• Option D (Add more S3 buckets): S3 buckets store source documents, not embeddings. Adding more buckets doesn't improve retrieval speed from the vector database where embeddings are indexed and searched.

Best Practice: For optimal Bedrock Knowledge Base performance, focus on:

1. Properly sizing OpenSearch Serverless capacity units based on workload
2. Optimizing index refresh intervals based on data freshness requirements
3. Ensuring appropriate shard configuration
4. Monitoring query performance metrics to identify bottlenecks