OMB Commands - Time comparision with OWB

Question

Hi
1.
I wrote a TCL script with OMB commands to Deploy the objects from my OWB to a destination environment.
The script is working fine but it is taking a bit more time than when deployed through OWB.
From OWB one module is taking 3.4 minutes but from the script it is taking 5.2 mins.
I am just trying to understand the reason why this script is taking more time.
I can think of the following reasons for this delay in the script.
a. Connect to the repository
b. Start the control center
My script is as follows.
if { [ catch {
OMBCREATE TRANSIENT DEPLOYMENT_ACTION_PLAN 'TREX_DEPLOY_PLAN' ADD ACTION 'TREX_DEPLOY_PLAN' \
SET PROPERTIES (OPERATION) VALUES ('REPLACE') SET REFERENCE SEQUENCE '$seqName'
OMBDEPLOY DEPLOYMENT_ACTION_PLAN 'TREX_DEPLOY_PLAN'
OMBDROP DEPLOYMENT_ACTION_PLAN 'TREX_DEPLOY_PLAN'
OMBCOMMIT } errmsg ] } {
Is there any other reason for this delay ? Or am I missing something ?
2.
Is there any difference between executing OMBCOMMIT 'once' in the script and executing it for each OMBDEPLOY ?
Which method is recommended ?
3.
We are planning to run the script in parallel for more than one module at a time.
Is there any way we can do this in TCL by getting the same connection and control center instead of initiating them for each thread ?
(Something like pooling of the connections and control centers ?)
I appreciate any help regarding the above doubts.
Thanks and Regards
Sameer

Answer 1

Hi Sameer,
1. There's no obvious reason why an OMB command should take any longer than the equivalent operation in the UI. The initial connection to the repository and the connection to the control center will take some time as you've mentioned, but probably not that long.
OMB*Plus doesn't support deploying a whole module in one go like the UI does, so i guess you have a TCL script that deploys the contents of the module as separate deployment plans? If this is the case, then this will introduce some overhard as each deployment will need to perform the same location checks and create the controller mechanisms for the deployment. If you haven't already done so, you could try adding all the objects from the module to a single deployment plan as this would reduce the overhead.
2. There shouldn't be any difference between doing commits per statement, or one big one at the end. I would just use one at the end.
3. OMB*Plus does have some support for parallel activities through the OMBDEPLOY ASYNCHRONOUS keyword. This will start the deployment and return to the command line so that you can continue with other tasks. Unfortunately though, you can't perform two deployments this way as only one generation can be performed at a time in each session.
In order to perform parallel modules, you'd need to use multiple TCL sessions, each of which would create their own distinct connections to the repos/ control center (there's no connection sharing across threads). Hopefully this shouldn't be a big overhead though.
Nigel.