Any mechanism used to push data out of the Data Lake can be defined as a Data Push mechanism. These can be of various types; however, here we are discussing some of the most common mechanisms:

• Data Exports: A serving layer must have the required tools, controls, and manageability to export the data in desired format for consumer applications. A part of this layer does indicate similar roles to that of an ETL; however, it is more driven by what the consumer application requires and how. These may be driven by internal batch processes scheduled at the serving layer such that the data from Data Lake can be extracted, transformed, and loaded to the destination. This part of the serving layer may also leverage some of the embedded ETL tools for the purpose.
• Data Publish: A serving layer may also publish the data to the required queues/topics for consumption by consumer applications as subscribers. This gets us to point-to-point data delivery and publish/subscribe models. Since these data had to be publishable messages, the size of these messages needs to be as small as possible for optimal performance.