Additional directories created on each engine tier computer
The following directories are created on engine tier computers. The directories are created by the InfoSphere® Information Server installation program.
- Project directories (runtime shadow repository) for the InfoSphere Information Server engine
- These directories will contain the following project-related items:
- Job executables
- Compiled routines
- Buildop stage types
- Job runtime artifacts
- Job status records
- Cached versions of design objects
- Message handler files
- DSParams file for project-specific variables that the IBM® InfoSphere DataStage® administrator defines
- Files that the Hashed File stage creates for server jobs (default location). However, a better practice is to create the files for the Hashed File stage on a separate file system.
Because disk space typically is limited in the default directory within the installation file system, you might want to create projects in their own file system. When a job is compiled, its source is extracted from the metadata repository and stored in the runtime repository to facilitate job compilation. Project directories can grow to contain thousands of files and subdirectories depending on the number of projects, jobs, and runtime metadata for each job.
InfoSphere Information Analyzer creates its own project directory for operations such as column analysis. In addition, each InfoSphere Information Analyzer project is associated with an InfoSphere DataStage project within InfoSphere Information Analyzer. The InfoSphere DataStage project directories are located on the engine tier at the location that is configured in InfoSphere DataStage.
The following table provides operating system-specific information about the project directories.
Table 1. Project directories: operating system-specific information Operating system Default directory path Additional requirements AIX® /opt/IBM/InformationServer/Server/Projects If the directory is on a network file system (NFS), export the file system with the no_root_squash option enabled and the nosetuid option off. Linux® /opt/IBM/InformationServer/Server/Projects If the directory is on a network file system (NFS), export the file system with the no_root_squash option enabled and the nosetuid option off. Windows C:\IBM\InformationServer\Server\Projects None - Data file storage directories
- These directories will contain temporary and intermediate data
files for the parallel engine:
- Engine temporary storage (scratch, temp, and buffer)
- InfoSphere Information Server artifacts, such as surrogate key files
- Segment files for parallel data sets
- Staging and archival storage for source files
The engine uses temporary storage to hold data sets and to temporarily buffer virtual data set records. Plan for these directories outside of the installation directory for InfoSphere Information Server.
Use local (internal) storage that is on a high-performance, shared file system with the same mount point that is available on all servers.
For better performance, create separate file systems for each data and scratch resource partition. For example, if the development and the production teams share an installation, separate the data and scratch directories for each team on different file systems to create separate environments.
After installation, edit the parallel configuration files to assign resources, such as the disk and scratch file systems, to jobs at run time. For more information about the parallel engine configuration file, see ../../com.ibm.swg.im.iis.ds.parjob.dev.doc/topics/c_deeref_The_Parallel_Engine_Configuration_File.html.
The following table provides operating system-specific information about data file storage directories.
Table 2. Data file storage directories: operating system-specific information Operating system Default directory path Additional requirements AIX - /opt/IBM/InformationServer/Server/Datasets
- /opt/IBM/InformationServer/Server/Scratch
If the directories are on a network file system (NFS), export the file system with the no_root_squash option enabled and the nosetuid option off. Linux - /opt/IBM/InformationServer/Server/Datasets
- /opt/IBM/InformationServer/Server/Scratch
If the directories are on a network file system (NFS), export the file system with the no_root_squash option enabled and the nosetuid option off. Windows - C:\IBM\InformationServer\Server\Datasets
- C:\IBM\InformationServer\Server\Scratch
None