If you are unable to create a new account, please email support@bspsoftware.com

 

News:

MetaManager - Administrative Tools for IBM Cognos
Pricing starting at $2,100
Download Now    Learn More

Main Menu

Information on No of threads used by job, Publish job etc?

Started by Preethi, 15 Oct 2010 02:13:54 AM

Previous topic - Next topic

Preethi

Hi All,

Where can i find the information about the Reconcile time, validation Job duration, Import Job, Publish time, cutdown time, admin links duration for all the applications.

I have to create a report which captures the information about Reconcile time, Publish job and no of threads used by them.

I know that i can go to application to get the above details. But we have around 110 applications, so its time consuming.

Is there any way i can get details from the tables?Can i automate the process?


Thanks and Regards,
P

ericlfg

Hi Preethi,

For applications:
My understanding of this is that each application will store these metrics until a new job is run in the P_JOB table in the application database.  Ex: A reconcile will run and present the metrics of this until a new reconcile is processed.  Also, depending on how much work is being done and what the reconcile was prompted for (synch with analyst, import data, admin link reconcile, etc) there may be a high fluctuation in processing time.   

For admin links and macro's:
General information can be gleaned from the P_INTERAPPLINK and P_MACRO tables in the planning store (or content store depending on your configuration) regarding start and end times.

As for the # of threads a job will utilize, this is easy to determine.  Each job server has a set number of processors, if you go through the monitoring objects, you can calculate how many threads will be utilized for each application, publish container, admin link, etc.

Ex: 2 job servers, 8 CPU's each.  1 cluster with both servers registered.  You will always (unless you modify the concurrent processors) have 16 threads running jobs for all applications, admin links, publishes, etc... if you have the objects monitored at the cluster level.  You can perform similar investigations based on your configuration.

So if you script a select statement that always appends and is formatted in a way that can isolate applications, publish containers, etc.. every 15 minutes or so, you may be able to get some metrics after a couple of days and use an average to configure a baseline.

Rutulian

Hi Preethi,

You can also run the Job Doctor at regular intervals - this interrogates the tables mentioned in the previous post and generates HTML output.  It's available as a Contributor Macro step.

It's pretty detailed, but not too configurable - if you need something very specific it's probably worth interrogating the tables, but this should give you a good idea of the kind of information you'll get out.