Friday, August 9, 2013

Distributed AD in Shared APPL_TOP

Today I've used disributed=y with adpatch option to allow workers to be started on remote machines, where they can utilize the resources on the remote machines when completing their assigned jobs.

Distributed AD offers improved scalability, performance, and resource utilization by allowing workers of the same AD session to be started on additional middle tier systems.

Prerequisites

1) Shared APPL_TOP
2) AD.H


Working

On one of your shared APPL_TOP nodes, start your AutoPatch or AD Administration session with the following command line options:

localworkers= workers=

For example to run an AutoPatch session with 32 workers on the local node and 32 workers on a remote node:

adpatch localworkers=32 workers=64

On one or more of the additional shared APPL_TOP nodes, start an AD Controller session with the following command line option:

adctrl distributed=y

After providing basic information, AD Controller will prompt for the worker number(s) to be started. For example, enter "33-64" to start workers 33 through 64. If AD Controller is started prior to AutoPatch or AD Administration starting the Jobs System, AD Controller will ask if you want to wait. Choosing yes will cause AD Controller to wait until the Jobs system is started, at which point it will start the appropriate worker processes. If an AutoPatch session has already been started, AD Controller will wait automatically.
Example of a two node session with 64 workers:

Node 1) adpatch localworkers=32 workers=64

Node 2) adctrl distributed=y and choose Enter the worker range 33-64


till FND_INSTALL_PROCESS tables get created, distributed adctrl sessions will wait for node1 adpatch. when fnd_install_processes table gets created, distributed workers will be forked and will start the tasks.

it's really cool way of reducing downtime - Part No. B31567-02 of R12 document set contains bit more information on this!!



So, Enjoy with more workers !!! ;)

1 comment: