Understanding SAS Grid Servers

Overview of SAS Grid Servers

The SAS Grid Server serves as a bridge between SAS applications and the grid environment. It enables an application to recognize the grid and submit jobs to it.

The grid server is actually a logical server, which is a component under a SAS application server. A grid consists of the following nodes:

a grid control server
a machine that distributes jobs to machines on the grid. A grid control server can also do work allocated to the grid.
one or more grid nodes
a machine or machines that run a portion of the work allocated to the grid.

A logical grid server is required on both types of nodes.

The grid control server and the grid node include the following components:

grid middleware provider
workspace server and spawner (grid control server only)
DATA step batch server
Base SAS
SAS/CONNECT server and spawner

The logical grid server definition specifies the command that is used by the middleware provider to start a SAS/CONNECT session.

To configure grid nodes, you first design the grid and designate a machine to be the grid control server. Next, you set up the logical grid server definitions on all of the grid nodes.

After a grid is configured, you can add grid nodes to increase grid capacity, create the required logical server definitions on a machine, and install the required software on each machine. Specific information about setting up and configuring a grid is on the SAS Scalability and Performance focus area: http://support.sas.com/rnd/scalability/grid/griddocs.html. If you are setting up a grid using middleware other than Platform Suite for SAS (such as United Devices GridMP or DataSynapse GridServer), specific values for the fields in a grid server definition are also on the SAS Scalability and Performance focus area.

Top of Page