A company uses Amazon S3 as a data lake and a multi-node Amazon Redshift cluster as a data warehouse. Data files in the lake are organized by their source, and the company currently issues a separate COPY command for each file location to load into a single Redshift table, which is slow. The company needs to speed up ingestion without increasing costs. What should they do?

Creating a manifest file is the most efficient solution because Amazon Redshift's COPY command can use a manifest file to load multiple data files from different S3 locations in a single operation. This significantly speeds up ingestion compared to issuing separate COPY commands for each file, without incurring additional costs. Using an Amazon EMR cluster or an AWS Glue job to consolidate files into one folder would add unnecessary processing and cost. Loading data into Amazon Aurora first and then using an AWS Glue job to move it to Redshift introduces an extra, complex, and costly step.