Guidelines For Model Design, Cube Creation, And System Maintenance Process; Setting Up Data Sources For Unix And Linux Cubes - IBM Cognos User Manual

When running Cognos Transformer on UNIX or Linux rather than Windows,
consider the following:
v Supported data sources include delimited and fixed-field text files, Cognos
v If you save your models as .mdl files, you can use an FTP program to transfer
v If you need to add security to your cubes, you can leverage any authentication
Guidelines for Model Design, Cube Creation, and System
Maintenance Process
Use the following guidelines when building PowerCubes in UNIX or
Linuxenvironments.
v Design and create a prototype model locally, on a Windows computer.
v Create and test your cubes locally on a Windows computer.
v Define a UNIX- or Linux-accessible data source in the model and save your
v Convert the .mdl files to .pyj files and create cubes using the .pyj files.
v Move your cubes to a UNIX or Linux production server and tune your system

Setting Up Data Sources for UNIX and Linux Cubes

Your models must be able to read data sources from the server where Cognos
Transformer is running. Because Cognos Transformer is intended for production
environments, it is likely that these data sources contain large volumes of data.
When you build a prototype, you must set up source files that provide access to a
subset of the data that is sufficient for designing your model.
The data sources you use to build the prototype must have the same columns as
those you intend to use later, in the UNIX or Linux component. The names of
columns used to define levels and measures in the models must match. However,
the columns can be in any order and may be unused. If there are fewer categories
in each column of the prototype source than in the production source, data records
192
IBM Cognos Transformer Version 10.1.1: User Guide
PowerHouse 4GL subfiles, IBM Cognos packages and reports, IBM Cognos
Impromptu Query Definition (.iqd) files (on IBM Cognos Series 7 version 7.4
supported platforms only) and other sources accessible through your database
connectivity software. To coordinate cube builds with your data warehouse
updates, run all related processes on UNIX/Linux servers.
them from your Windows modeling computer to your UNIX or Linux
production environment. You can then use the processing capability of your
more powerful UNIX or Linux servers, to quickly and efficiently create cubes
from one or more models, tuning your system to avoid network bottlenecks. You
can run Cognos Transformer from the command line or by using Model
Definition Language (MDL) scripts. Client-server modeling is not supported in
Cognos Transformer.
Tip: After transferring your .mdl files to UNIX, you may want to convert them
to .pyj files because .pyj files load faster than .mdl files. The command line
syntax for the conversion is as follows:
cogtr
-m model.mdl -s model.pyj
provider configured in IBM Cognos BI, apply custom views in Cognos
Transformer for applicable users, roles, and groups, and deploy the resulting
cubes to your UNIX or Linux environment. For more information about setting
up and administering secure IBM Cognos deployments, see the Administration
and Security Guide and the Architecture and Deployment Guide.
server cubes in Model Definition Language (MDL) format.
for optimal OLAP reporting performance.