Distributed Batch Processing

Lawson supports distributed processing of batch jobs. This means that, as a way to improve system performance, batch processing can be distributed across more than one physical machine. Communication between servers is via TCP/IP sockets. The server that runs LAJS is the master but client requests can come from either machine. In a remote processing scenario, two job execution programs (tcpexecjob and tcpexecjobd) are called in addition to execjob.

A job can be submitted from any machine. LAJS determines if the job is configured to run locally or remotely. If the job is to be run locally, EXECJOB starts.

If the job is remote, the operating system or the remote machine (as determined by inetd, the operating system daemon that listens for port activity) starts TCPEXECJOB. TCPEXECJOB creates a connection with the subordinate machine and then sends the job to TCPEXECJOBD for processing. TCPEXECJOBD executes the job on the remote server.

Print and log files must reside on a shared resource when distributed processing is used.

The following diagram is a visual rendering of the distributed batch process. The drawing details with communication only. It does not show details about processing. Except for the invocation of special distributed batch programs, the execution cycle for distributed batch processing is similar to that of single-server installations.