This post was kindly contributed by Business Intelligence Notes for SAS® BI Users - go there to comment and to read the full post. |
Collecting ARM logs can be extremely useful when tuning OLAP cubes for faster response. The ARM log (which stands for Application Response Measurement) collects vital usage information for OLAP cubes using the SAS OLAP Server logging facility. SAS OLAP Cube studio can analyze these logs to assist with aggregation building to make OLAP cubes return faster results. Angela Hall discusses this and other ways to tune OLAP cubes in this post. SAS Support has excellent information on tuning OLAP cubes using ARM logs as well.
Enable ARM Logging Interactively
You can start collecting ARM logging by setting a value in SAS Management Console.
By default the log is placed in the following location on the server:
[SAS CONFIG]/Lev1/SASApp/OLAPServer/Logs
Enable ARM Logging Automatically
When developing OLAP cubes I use ARM logging to build key aggregations I find important during testing or to ensure the initial performance is acceptable for end users. Since a development environment can go up and down (especially on my own laptop ) I find myself continually going into SAS Management Console to enable ARM logging. Rather than going through this manual approach every time the OLAP server is restarted, I recommend enabling ARM logging by default in development environments.
To enable ARM logging automatically when the OLAP server starts, find the logconfig.xml file located in the following location on the server:
[SAS CONFIG]/Lev1/SASApp/OLAPServer
Within the logconfig.xml file simply find the ‘Perf.ARM.OLAP_SERVER’ option and change the ‘value’ option to “Info”:
Note: ARM logging can fill up the server over time if they are not cleaned up. Enabling ARM logging using this technique will ALWAYS produce logs as long as the server is operational and OLAP cubes are accessed!
Photo credit: MS Office Clipart
This post was kindly contributed by Business Intelligence Notes for SAS® BI Users - go there to comment and to read the full post. |