SAS/CONNECT User's Guide

Agent Scheduling

Agent services is used with compute services and messaging services to provide client/server based task management for the nodes across your network. An agent is simply SAS source code that is launched by the DOMAIN server to support the execution of a task on a remote node. After the task has finished executing, an agent may be designed to send a completion notification to a message queue for the client application, or the client can check the DOMAIN server to see if the agent has completed execution.

The source that composes an agent may be pre-defined and stored with the DOMAIN server, or the source may be defined dynamically with the request for the agent to execute.

Client/server-based agents provide the following types of services:

distributed agent processing
periodic agent processing
conditional agent processing
parallel agents processing.

These services can be used together to maximize the flexibility and functionality of a client/server application. For example, using agent services, an application can be designed to execute autonomously on a periodic basis. The application can make conditional decisions to satisfy its goals, including submitting other processes for execution on various hosts on the network. See Using Agent Services for more information.

The AGENT option must be specified in the PROC DOMAIN statement, in order to use agent services. An example follows:

libname domain ".";
proc domain agent;
run;

For more information about configuring and using the DOMAIN server, see Communications Access Methods for SAS/CONNECT and SAS/SHARE Software.

Chapter Contents
Previous
Next
Top of Page