OEM Enterprise Manager

[naushi]

I am going to be creating an OEM repository and monitoring some servers. My task is to report some servers for UP and DOWN status, number of users threshold reached etc.

I though have no problem building all this but dont know how to accumulate the status or the results and build some kind of graph (trend) to show e.g what is the down time for a specific server, how many times a certian batch job failed or succeeded.

Is any one familiar with how to build the trend in a file or something. Is there any tool that you may be familiar of that I can buy that might help build these trends.

This is an important task for us and I would appreicate any help.

Regards,
Naushi Hussain
Naushi.hussain@honeywell.com

 

[JtheRipper]

Hi,

I personally only find OEM helpful when you want to know something NOW, not for keeping history. I had the same type of problem, but I created a few tables and a procedure that captures these type of information and stores it in the tables. I run the procedure every 5 mintutes and it works like a charm. I would suggest you do the same because if you first have the data you will be able to manipulate/draw charts/etc with it whichever way you like.

 

[pscdba]

There is a history feature in OEM but it is limited. A very good product for monitoring the OS and the RDBMS is Patrol from BMC Software. It is efficient, stable and provides a very good base for trend analysis. The product does not consume a large amount of resources. It includes the History Propogator to copy the trend data to a centralized history directory on a single server and a History Loader which loads the history data into any of several target RDBMS products if you desire. (You can analyze directly from the files if you wish.) We have been using the Patrol Knowledge Module for Oracle and the Patrol KM for Ingres for years and found them to provide many parameters which are useful for both trend analysis and troubleshooting. The product architecture is very flexible and reconfiguring thresholds is a one minute process. Creating and setting up new parameters is very reasonable. Simple ones can be done in about 10 minutes if you have a script which generates the desired result value. More complex parameters and automated recovery actions take considerably longer but do not require a high level of expertise in Patrol. Monitoring some OS parameters with the same tool allows you to compare effects on or from the OS with the same polling period in the DBMS. Technical support from BMC has been very good although their admin functions have been a pain.