1. Greenplum initiates read request from Greenplum with Hadoop NameNode
having access to the required file blocks.
2. Hadoop NameNode sends instruction to DataNode to provide file blocks to
Greenplum thus addressing the read request.
Following are the steps to write data to Hadoop HDFS from Greenplum Database:
1. Greenplum (client) initiates write request with Hadoop NameNode having ac-
cess to the required file blocks.
2. Hadoop NameNode sends instruction to DataNode to update the file blocks
into the file including the instruction for any data replication.
In the newer versions of UAP, there are new integration frameworks being built,
like HAWQ, that provide SQL-like querying capabilities over HD and can query
Greenplum Database and HD combined to provide the client with a single shot
view on data across Greenplum Database and HD.