Is it possible to create a partitioned table in AWS Glue Data Catalog over AWS S3 files and have a different data schemas in different partitions of this table?

If so, will I be able to create this partitioned table as an external table in Redshift with help of Redshift Spectrum and query it?

CodePudding user response：

If the files are of json/parquet type, then define the table such that it includes all the columns - some columns may be present in some partitioned files and others in other partitioned files.

If the files are of csv type, then you need to have the columns present in the same order in each file but each file may not include all the possible columns.