A company reads data from multiple Amazon RDS customer databases where field names and formats are inconsistent (for example, place_id in one database and location_id in another). The business must link customer records across databases even when field names differ. Which solution provides this capability with the LEAST operational overhead?

The correct solution leverages AWS Glue's FindMatches transform, which is specifically designed for identifying duplicate or matching records in datasets with inconsistent schemas, even when field names differ. This is a managed service, significantly reducing operational overhead compared to self-managing EMR clusters or building custom Spark ML pipelines. An AWS Glue crawler automatically discovers schema information from the RDS databases, populating the Glue Data Catalog, which is a prerequisite for using FindMatches. The other options involve more operational overhead: EMR clusters require provisioning and management, and building custom Spark ML models in SageMaker or EMR demands more development effort and expertise for data matching, especially with schema inconsistencies.