Page 1 of 1

Regarding Information Analyzer jobs on Multiple nodes

Posted: Tue Dec 07, 2010 2:45 am
by arvind_ds
Hi Experts,

I have a small question regarding Information Analyzer 8.1.

Is it possible to run IA jobs(column analysis, primary key analysis, foreign key analysis etc...) across multiple engines.?

For example, I have the below setup.

(1) Client tier : Computer A
(2) Services tier along with colocated engine tier: Computer B
(3) Separate Engine tier : Computer C

Is it possible to run IA jobs across Computer B and Computer C to take advantage of aditional engine capabilities which is on Coputer C in this case.?

OR

IA job can be executed through one engine at a time only.?

Please share your thoughts on this.

Posted: Tue Dec 07, 2010 3:35 pm
by ray.wurlod
IA jobs run as parallel jobs. Therefore the answer is "yes". Even better, apart from providing a valid configuration file, you don't have to do anything to take advantage of this functionality.

Posted: Wed Dec 08, 2010 5:29 am
by arvind_ds
Thanks for your quick response Ray. Do I need to register the aditional node by using Information Analyzer --> Configuration --> Analysis Settings --> Validate engine connection options for all the nodes and then update default.apt configuration file with all nodes.

Actually I tried to do this and while validating separate engine connection, i got the below error.

Analysis Connector:Error
Invalid settings:
Error executing DS jon:

Posted: Wed Dec 08, 2010 1:27 pm
by ray.wurlod
No, simply mention a second node in the DataStage configuration file (e.g. default.apt) on the analysis engine. Validate the analysis engine - the one on which the ANALYZERPROJECT project exists. You do have to do all the deployment things that parallel engine requires - sharing the parallel engine software and enabling rsh/ssh between the nodes.

Posted: Thu Dec 09, 2010 2:53 am
by arvind_ds
Thanks Ray. I have one more question. Should I NFS mount the Projects directory from conductor node onto computational node such that the Projects(ANALYZERPROJECT and other Projects...) are visible through all the engines.?

If I don't do this, I get error message saying file or directory(/opt/IBM/InformationServer/Server/Projects/ANALYZERPROJECT) does not exist on computaional node.

Please share your inputs on this.

Posted: Thu Dec 09, 2010 4:05 am
by ray.wurlod
Your middle paragraph says it all.

Posted: Thu Dec 09, 2010 4:27 am
by arvind_ds
That means I have to NFS mount the Projects directory across all the nodes such that the Projects are visible across all the nodes, as I understood from your last post.Please confirm.

Posted: Thu Dec 09, 2010 3:44 pm
by ray.wurlod
It does. And the PXEngine too.

Posted: Fri Dec 10, 2010 1:15 am
by arvind_ds
Thank You So Much Ray.