Database Reference
In-Depth Information
)
;
Figure 10.10
Simple Polybase query against data held in Hadoop
PDW is able to use the external table
HDFS_FactInternetSales
to read
through to Hadoop and import the data residing in HDFS using a DMS
operation called an
ExternalRoundRobinOperation
. This operation
imports the buffers as is and simply allocates them on a round-robin basis
to the compute nodes. As we do not need to join this table to any others,
there is no need for anything more complex. Once allocated, the buffer is
simply bulk inserted into a temporary table where it can be easily read by
the
ReturnOperation
as shown in step 3 of the plan. The results are then
streamed back to the client that issued the original query.
PDW can also initiate joins and aggregations between data held in one
of its tables and data held in HDFS. In the following example, PDW is
able to read data in HDFS through the same external table
HDFS_FactInternetSales
andsubsequently join ittoanumber oftables
in PDW (see
Figure 10.11
):
