Polyglot is a very common word in computing, which means, multiple.

In the serving layer there will be multiple data stores (persistence mechanism) being used to store the same data in a variety of models as dictated by the use cases. We really don't know whether there is a term like Polyglot data access in the industry. By polyglot data access, we mean, existence of multiple data stores in the serving layer, which are being used by serving layer to churn various data services to the end users.

In our Data Lake we already have polyglot data access in action by using Elasticsearch, Hive and HBase as the data access stores.

The concept of Polyglot access can be further extended into persisting parts of information held by an object into multiple target data stores and access them as well. There are frameworks today support such mechanisms for Big Data technologies as well.

One of the very prominent use cases of polyglot access has been to to store the indexes and the data separately, such that the indexing technologies are primarily used for indexing and data technologies are primarily used for efficient storage. There a number of options for polyglot persistence frameworks in context of Big Data, with Hibernate like JPA interface to enable developers to define objects using standard persistence configurations and annotations.