The LASR procedure can
transfer data from any data source that SAS can read and load it into
memory on the SAS LASR Analytic Server. However, the LASR
procedure can also be used to make the server read data from a co-located
data provider. The HDFS that is part of SAS High-Performance Deployment of Hadoop
provides a co-located data provider. Some third-party vendor databases
can also act as co-located data providers. Two examples of third-party
vendor databases are the Greenplum Data Computing Appliance (DCA)
and Teradata Data Warehouse Appliance. When the data is co-located,
each machine that is used by the server instance reads the portion
of the data that is local. Because the read is local and because the
machines read in parallel, very large tables are read quickly.
In order to use a third-party
vendor database as a co-located data provider, the client machine
must be configured with the native database client software and the
SAS/ACCESS Interface software for the database. The database is identified
in a LIBNAME statement. The LASR procedure then uses the SERVER= information
from the LIBNAME statement and the host name information in the PERFORMANCE
statement to determine whether the data is co-located. If the host
information is the same, then the data is read in parallel.