AWS offers no less than six different cloud database and SQL/NoSQL services: RDS, Aurora, DynamoDB, Redshift, Athena, and AWS Database Migration Service! Out of all these services, only two are compatible with Amazon Machine Learning: RDS and Redshift. You can store data in either service and create datasources from these sources. The datasource creation methods for the two services have similar parameters, but differ quite significantly when it comes to the underlying AWS service communication.
RDS and Redshift are very different services. Redshift is a data warehouse used to answer a few complex and long running queries on large datasets, while RDS is made for frequent, small, and fast queries. Redshift is more suited for massive parallel processing to perform operations on millions of rows of data with minimal latency, while RDS offers a server instance that runs a given database. RDS offers several different database types – MySQL, PostgreSQL, MariaDB...