SAS 9.4 Language Reference, 6th Edition

SAS Grid Manager

SAS Grid Manager provides scalable workload management and prioritization, high-

availability processing, and optimized performance across various SAS processes and

services that execute in threads on a grid of configured servers. By coordinating the

resources of separate servers into a centrally managed SAS environment, SAS Grid

Manager can provide accelerated processing for SAS jobs.

SAS programs, that are grid-enabled to run in multiple independent steps within the

overall program flow, execute in threads. The threaded execution typically occurs at the

DATA step or procedure boundaries. Because the programs are specifically written to

execute in threads, it is assumed that the THREADS option is set and the CPUCOUNT

option is greater than one. SAS Grid Manager enables subtasks of individual SAS jobs to

run in parallel on different parts of the grid and share a pool of resources. The threaded

subtasks can further benefit from threaded processing performed by any of the thread-

enabled procedures executed from within the code. Programs that have been analyzed

for parallel processing using the SAS Code Analyzer can be run on the grid with SAS

Grid Manager. This automatically assigns the identified subtasks to a grid node.

Many SAS products, such as SAS Enterprise Guide, SAS Enterprise Miner, SAS Data

Integration Studio, SAS Web Report Studio, SAS Marketing Automation, and SAS

Marketing Optimization are integrated with SAS Grid Manager. An option within the

application or in the SAS Metadata enables the integration. SAS Data Integration Studio

and SAS Enterprise Miner have code generation engines that can recognize opportunities

for parallelization and generate the appropriate code to submit to SAS Grid Manager to

execute in parallel across the grid.

Other SAS components, such as the SAS Intelligence Platform, use SAS Grid Manager

to determine the optimal SAS server for processing by distributing server workloads

across multiple computers on a network. SAS Grid Manager divides jobs into separate

processes that run in parallel across multiple servers. SAS Grid Manager is documented

in Grid Computing for SAS.

SAS In-Database Technology

SAS In-Database technology provides faster execution of commands and functions by

sending them to execute inside the databases by SAS applications. This process avoids

data movement and conversion but also takes advantage of the threading capabilities of

the host database.

SAS In-Database runs on a variety of threaded DBMS and hardware or software

appliance architectures such as Teradata, Greenplum, Aster Data, Netezza, and DB2.

Because the threading is controlled by the database, NOTHREADS and CPUCOUNT

options have no effect.

In-Database processes are divided into multiple parallel tasks within the database, each

working with small subsets of the overall data to be processed. As the results of the

smaller subsets are derived, they are consolidated and returned to the SAS application.

Typically, th execution times are much shorter than if the data were transferred to the

SAS server.

In-Database technology uses existing database functionality and standard SAS

functionality to extend that SAS functionality into the database. For example, Base SAS

SAS In-Database Technology 215