Denormalization or CDM flattening stops when processing large amount of data

When you upgrade from CMDB 8.1.02 to any later version, you have to denormalize or flatten the common data model to improve the performance of the CMDB. The denormalization process is performed using the CDM flattening utility.

Issue when performing denormalization

When running the CDM flattening utility on a very large number of configuration items (CIs), sometimes, the migration may stop. This is because the utility is configured by default to run on multiple threads for faster performance but at the cost of redo space on the Oracle server. Sometimes, this can result in the server running out of redo space when a very large quantities of data involving millions of CIs is denormalized.

Resolution

You must change the flattening utility settings by performing the following steps in Remedy AR System Administration Console. 

1. In a browser, enter the URL designated by an administrator. For example:
   http://<midTierServer>:<port>/arsys
   <midTierServer> is the name of the web server where the Remedy Mid Tier is installed*.*
   <port> is the port number of the web server where the Remedy Mid Tier is installed, which is required only if the web server is not using the default port (80 if using Microsoft IIS or 8080 if using Apache Tomcat).
2. In the login page, specify the following details:
   • In the User Name field, type your user name.
   • In the Password field, type your password.
   • In the Authentication field, type the authentication string (if one was provided by an administrator)
3. Click Login.
4. In the Remedy AR System Home page, go to AR System Administration > AR System Administration Console > System > General > Centralized Configuration.
   
   
5. Select the component com.bmc.cmdb.server.shared.
   
   The CDM-Flattening-Disable-Parallel-Processing property is displayed with its default value.
   

6. Change the value of CDM-Flattening-Disable-Parallel-Processing to True.
   Now the CDM flattening utility runs with smaller redo space requirement and does not stop when processing large amounts of data.

Related topics

Denormalization of the Common Data Model