A company has 10 TB of data in an Amazon Redshift cluster. The data engineering team uses SageMaker Studio for analysis and model development, but only a subset of the Redshift data is needed for training. They need a secure, cost-effective way to export the relevant subset to Amazon S3 for model development. Which solutions satisfy these requirements? (Choose two.)

The correct options provide secure and cost-effective ways to extract a subset of Redshift data. Using a SageMaker Studio notebook with a Python database adapter and credentials from Secrets Manager is a standard, flexible, and cost-effective approach for querying and exporting data. SageMaker Data Wrangler directly integrates with Redshift, offering a visual interface for data preparation, including querying, transforming, and exporting to S3, which is efficient and user-friendly. Incorrect options: A SageMaker Processing job with Spark is overkill for simply querying and exporting a subset, incurring unnecessary cost and complexity compared to the direct methods. Downloading 10 TB of data to several medium notebook instances is impractical and inefficient; notebook instances are not designed for such large-scale data processing or storage. Launching an extra-large SageMaker notebook instance with 10 TB block storage to download all data is extremely expensive and inefficient. Notebook instances are not meant to store or process entire large datasets locally.