A financial services firm is building a serverless data lake on Amazon S3 that must: allow querying old and new S3 data with Amazon Athena and Redshift Spectrum, support event-driven ETL pipelines, and provide easy metadata discovery. Which solution meets these requirements?

This solution effectively addresses all requirements. An AWS Glue crawler automatically discovers schema and partitions for new and old S3 data, populating the AWS Glue Data Catalog, which Athena and Redshift Spectrum can query. AWS Lambda can trigger AWS Glue ETL jobs in an event-driven manner (e.g., upon new S3 object creation), processing data as needed. The AWS Glue Data Catalog provides centralized metadata discovery. Incorrect options: Using AWS Batch for ETL adds unnecessary complexity compared to Glue ETL for serverless data lake processing and an external Hive metastore complicates metadata management. Triggering AWS Batch jobs via CloudWatch alarms is less event-driven than Lambda for S3 events and AWS Batch is not ideal for serverless ETL in this context. An external Apache Hive metastore introduces management overhead and is not as integrated as the AWS Glue Data Catalog for AWS services.