There are some technical benefits of Spring Data over Spring + JPA, which in a pure SQL environment, I think give Spring Data an advantage:
- Spring Data uses the same CrudRepository interface for all implementations, so you'll have less effort to switch between JPA to MongoDB
- Spring Data saves you writing the same methods again and again. You just add the method to the interface and it'll generate it for you (e.g. UserRepository.findByUsername())
- You can save boilerplate on REST implementations for JPA, MongoDB and others (see http://projects.spring.io/spring-data-rest/)
- If you wanted to experiment with other persistence or indexing services, then there are Spring Data implementations for both mature and newer technologies such as for Neo4j, Hadoop, Solr, ElasticSearch, fuzzydb.
Given that you use MySQL and MongoDB, I think Spring Data is a strong candidate, as it allows developers to code to a single data access API (Spring Data) instead of two (JPA and the MongoDB Java Client).
Regarding the existing architecture, it sounds as though your manager layer is implementing either a Rich Domain pattern, or Active Record.
Spring Data is in my view very well suited to Rich Domain when combined with injection of services using Spring's @Configurable.
Lastly, I'd say that Spring Data also gives a significant advantage when needing to implement services for things like Spring Security and Spring Social, which use MongoDB or others instead of SQL.
We did this in the fuzzydb sample webapp that can be found here. (Disclaimer: I'm the currently sole recent committer on fuzzydb, and haven't touched it for a number of years, but we did have a live service, www.fridgemountain.com, based on that code, but neglected to promote it)