Memory Usage and IBM Planning Analytics

IBM Planning Analytics (PA) is an advanced analytics solution that uses an in-memory database architecture to produce real-time results. But what is an in-memory database? An in-memory database (or IMBD) is on that stores data in the computer’s main memory, instead of a hard disk drive, allowing faster response times because there is no need to access a disk to query or update data.

How it Works:

When a Planning Analytics (or TM1) database starts up, the TM1 server service (tm1sd.exe) reads data, cubes, dimensions, and other system information for a specific database from the disk and loads it into memory making it available to its users (note you will have a tm1sd.exe for each Planning Analytics database that you have).

The server service (tm1sd.exe) itself requires computer memory to run but the specific Planning Analytics database also consumes memory and, depending upon how complex the database is, it will naturally consume higher levels of memory (there are many other reasons that affect your database memory consumption, but note that a databases memory usage does not relate to the size of its “out of memory” data files saved to disk).

A helpful technote that explains that the memory consumed by the database is not the same as the memory consumed by the Windows process (tm1sd.exe) can be found here:

https://www.ibm.com/support/pages/memory-usage-value-inside-planning-analytics-administration-paa-website-seems-too-high-caused-misunderstanding-what-value-actually-measures

During routine use of a Planning Analytics database, “actions and events” – views, calculations, and other types of processing will temporarily allocate additional memory, and, IBM reports, that Planning Analytics does not “release” memory back to the operating system until the server is stopped or restarted, instead, it puts memory it no longer needs into a “garbage list” (to be reused as required). Therefore, the memory consumption (for the TM1 server/instance), as reported by the operating system, is the total of:

Actual Current Memory In-Use + Garbage-Marked Memory

Checking on Memory using PAW

IBM Planning Analytics Administration includes the ability to monitor and administer your databases from the Administration page (note that you must be an administrator to see the Administration tile on the Planning Analytics Workspace home page and to use the Administration page). Open the Administration page by clicking the Administration tile on the home page, then click the Databases tile which will show the total number of databases in your environment, along with a status summary for all databases. From there you can click any database name on the Databases list to view details for the database, such as its “overall heath” which includes current memory consumption statistics.

Another Option

To get an accurate report of the actual memory usage and garbage memory, you can view the }StatsForServer control cube, which tracks performance statistics for the server, including the number of cell updates, retrievals, view calculations, and retrievals for each connected client, active threads, and memory usage. The }StatsForServer control

Cube is one of the performance monitoring control cubes that you can enable.

More Information

For more information on the above information of if your ready for a deeper discussion on Planning Analytics memory consumption, contact QueBIT, we’d ready to help!