Most software engineering practioners that use Spring will be aware of the power of abstraction that Spring Data can provides to greatly reduce time taken and the amount of code necessary to interact with an application’s database. Most will probably also realize that the abstraction provides a degree of freedom in shifting to different data sources with minimal amount of code changes (generally importing a different dependency and potentially some modifications to the entity/POJO classes that are used to define the structure in the database). But what about using multiple Spring Data data sources in the same application that reuse much of the entity/POJO classes to efficiently provide access to different databases that serve different business use cases, such as fast paths of data access for hot data and slower paths for historical data? Maybe we need to store recent data in a database with faster access, and older data in a cheaper (slower) database (and maybe the database we select for our fast access doesn’t have the ability to run on cheaper, slower spinning disks). Using Spring Data Spanner (using an underlying JPA implementation) and Spring Data JDBC with a JDBC library for GCP BigQuery, we can serve up a consistent interface to our application (latency witholding) to our application data while preserving all the great benefits we get from Spring Data, as we’ll see below.