Posted by Dylan Wan on July 16, 2015
An ODI agent is a process running for performing the data integration jobs.
The data integration jobs are ODI scenarios in ODI. Each scenario can be regarded as an executable program.
However, we do not run those ODI scenarios directly. These jobs are executed by the ODI agent.
An ODI agent is a Java program. It connects to an ODI repository to get those scenarios to be executed. It also use the repository for writing the logs. You need to provide the information about which ODI repository it connects to, in order to start the ODI agent process. On the machine you are running the ODI agent, you do not need to have the full ODI installation. Only the Java library and Jar file that includes the ODI agent is needed. ODI studio software is not needed.
Since it is a Java program, you need Java runtime installed. Since it connects to ODI repository database, JDBC is required. Since the technologies used by scenario varies. The machine that the ODI agent is running needs to have those technologies available.
An ODI agent is a server process. It is allocated with a PORT number and after its starts, it is listening the requests. This means that it can accept the remote request. In the other word, submission of the job may come from a remote machine.
Typically you will submit the job from the ODI studio. You can also submit the job from ODI SDK.
During the design time, you do not need to run a standalone ODI agent. An ODI agent is embedded within the ODI studio. The job submitted within the ODI studio can be executed by the local agent.