After you launch Performance Modeler, the “Model Design” pane displays to suggest the steps you need to take to create a model. There are 4 steps, including:
- Defining dimensions
- Building cubes
- Creating “links” between your cubes and finally,
- Adding Rules and Processes.
Defining Dimensions
Using TM1 Architect you can select Dimensions, right-click and select Create New Dimension. Similarly in Modeler, you can select almost any folder shown in the Navigation Pane, right-click, select New and then “Dimension”.
When you create a new dimension using Performance Modeler, you have to tell Modeler what type of dimension you want to create (this is a little different from what we have in TM1). Fortunately, Performance Modeler gives us the Create new dimension dialog where you can name your dimension and select a type (Generic, Calculation, Hierarchy, Time or Versions).
Here is a high-level description of the dimension types (you can check the help file for more details). Something to keep in mind – if you are not sure of what type your new dimension should be, choose “generic” (Performance Modeler gives you the ability to change a dimensions type later).
Calculation dimensions
Contains formulas that perform mathematical and other operations on your data.
Time dimensions
Contains time members (of course) that are meaningful to your users, such as financial accounting periods or the dates of sales transactions.
Versions dimensions
Contains data from various iterations of a member in an application.
Hierarchy dimensions
Contains a representation of the reporting structure of your business, department, or enterprise.
Generic dimensions
Contains general members, such as lists of departments, products, or customers.
Simplest Example
So once you’ve entered a dimension name and select a type (in my example I named the dimension “Product” and set it to “Generic”), Modeler will display the dimension editor. Across the top, you have the ability to:
- Type in an element name
- Set up the elements format
- Setup a Pick List for element
- Create an N Level rule (calculation) for the element
- Create a C level rule (calculation) for the element
- Provide an element weight
Along the bottom of the dimension editor, you will see (depending on what you selected above) either basic properties information or options to select/edit or a mini “rules editor” to easily define your element rules (calculations).
Once you’ve completed editing your dimensions content, you’ll need to click SAVE to actually commit your new dimension and its content to the server.
Something to be aware of
While exploring the features and functions of Performance Modeler, I created several “test” dimensions of various types and with different content. In another step (which I’ll blog next about) I used these dimensions to create a new cube. I played with the cube (creating several views and subsets) and loaded some data. When I was done, I deleted the cube and the dimensions and closed modeler.
Later, on a flight to Phoenix, I felt the urge to do some more with Modeler, so I fired up the laptop and launched Performance Modeler. I received the following message:
After a restarting (all) my TM1 services and even a (desperate) rebooting of my machine I finally discovered the issue – invalid views and/or subsets in the TM1 database. During my previous session I had created and saved several views and subsets with the (now) non-existent cubes and dimensions. It seems that Performance Modeler doesn’t like this (where as TM1 Architect usually will tolerate this situation by simply ignoring them).
IBM recommends that you (try):
- Open Architect and open every view and subset – if something doesn’t open then delete it
- Find suspect views and subsets on the hard driver and physically delete them
- Stop and restart the TM1 services
If this doesn’t work (and it didn’t for me) then verify that you have Cognos TM1 10.1 Fix Pack 1 installed. In this installation a fix was applied to Cognos TM1 Performance Modeler to display a more detailed dialog titled “Error loading objects” which reports the names of corrupt objects upon startup and (and this is the important part) in this version, Cognos TM1 Performance Modeler opens, but does not render the corrupt objects.
After performing this installation, Modeler was happy to work with me again. Lesson learned though – be careful when removing objects previously created with Performance Modelers.
Another note, if you are on LinkedIn, please check out (and hopefully join) my new group:
Cognos TM1 Performance Modelers
Where I’ll be sharing best practices, hints and tricks and great examples for using PM.
Cheers!
