TM1 Server updates

IBM® Planning Analytics Local version 2.0.5 and the cloud release of IBM Planning Analytics version 2.0.5 includes the following features for IBM TM1® Server version 11.3.0.

Change in the behavior of the STR function

The behavior of the STR function has been updated. This may cause a behavior change in usage depending on which version of IBM Planning Analytics you are upgrading from.

For more information, see STR.

Change in the behavior of TI process rollback

A TurboIntegrator process that encounters a major error now rolls back instead of committing any changes.

This change in behavior impacts only major errors. The entire TurboIntegrator process is a single atomic transaction. Either the entire process succeeds and is committed, or the entire process fails and rolls back. For example, errors such as the data source failing to open or attempting to write to a cube that doesn't exist are major errors and now trigger a rollback.

The handling of minor errors is unchanged. For example, errors such as having a malformed input line in a data source logs a minor error and continues running the next data source record.

Multi-threaded loading of individual cubes

You can enable multi-threaded loading of individual cubes, which delivers better speeds than the approach that is enabled by the MaximumCubeLoadThreads setting. The following tm1s.cfg settings are available to enable and configure this feature:

  • MTCubeLoad
  • MTCubeLoad.MinFileSize
  • MTCubeLoad.UseBookmarkFiles
  • MTCubeLoad.Weight
  • IndexStoreDirectory
  • PreallocatedMemory.BeforeLoad
  • PreallocatedMemory.Size
  • PreallocatedMemory.ThreadNumber

Using MTCubeLoad also eliminates the risk of changing feeder generation sequences. When MTCubeLoad is enabled, the MaximumCubeLoadThreads configuration option is ignored.

TM1 Server administrators can use new tm1s.cfg settings (PreallocatedMemory.Size, PreallocatedMemory.ThreadNumber, and PreallocatedMemory.BeforeLoad) to configure preallocation memory and optimize scale-up and performance results.

The following loggers have been added:
log4j.logger.TM1.Server.Loading=DEBUG
Allows you to capture individual timing of TM1 Server loading stages: DeSerializeDimensions, DeSerializeAttributeCubes, DeSerializeRegularCubes, GenerateServerSecurity.
log4j.logger.TM1.Cube.Loading=DEBUG
Captures cube loading time, separately for .cub and .feeder files.

For more information, see MTCubeLoad.

New operational administrator

IBM Planning Analytics has a new system's user group, OperationsAdmin, whose sole purpose is to do maintenance and operational work. The members of this group do not see any of TM1 metadata, nor any cube or any other data. However, members of this group can perform tasks such as unlock users, disconnect users, and cancel threads.

For more information, see Administrative groups and authority.

Reference a public subset by name in MDX

You can use the TM1SubsetToSet function to return the members of a public TM1 subset. If a private and a public subset have identical names, enter the optional scope parameter as public to return the members of the public TM1 subset. If the scope parameter is not specified, the server searches first for a private subset.