Load from Table

SAS Visual Analytics can load tables into memory on a SAS LASR Analytic Server from a table that is registered in SAS metadata. If the table is from a DBMS library that is co-located with the SAS LASR Analytic Server instance, then the server instance recognizes that the table is from a co-located data provider and reads the data into memory in parallel. If the table is not from a co-located data provider (for example, a SAS data set), then the table is read from the data source, transferred to the root node on the cluster, distributed to the machines on the cluster, and then loaded into memory.

To load from a table:

Locate the table using the SAS Folders view or by searching. Right-click the table icon and select Load from Table.
On the General page, specify values for the Job name, Location, and Description if you intend to save this action as a job.

Select the Settings page and specify the SAS LASR Analytic Server and data parameters:

Field	Description
Configuration directory	This field specifies the location of the server description files.
Server description file	This field specifies the name of the server description file.
Table	The table name is populated automatically.
Table description	Specify a description for the data. This description overrides the description that is stored with the table. The description is displayed beside the table name in the explorer interface.

Click Submit.

The following items provide additional information about loading from a table:

The table to use as the data source must be registered in metadata before you can perform this task.
For deployments that use a co-located data provider, the SAS LASR Analytic Server performs a comparison of IP address to determine whether a parallel read is possible. The IP address that is in the SERVER= option of the generated LIBNAME statement must match the IP address of the SAS LASR Analytic Server root node.
When the data is loaded from a data source that is not a co-located data provider, it is transferred to the root node of the server, distributed evenly among the machines in the cluster, and then loaded into memory.