Data repository layer

The Data repository layer may be composed of multiple types of databases. The reason for this is that different data types may need to be represented for different purposes and usually a single choice of database would not suffice. The idea behind the serving layer is that it should be able to serve the data as per the requirement, hence multiple databases do find a good use here. In general the serving layer would serve a combined view of data both from the batch layer as well as from the speed layer.