Tm1 "wizard” approach

[ravi]

Hi All,

Just thought to get some comments on the-Learning tm1 using the “wizard” approach, when comes to IT process creating.

Has anybody have built tm1 model in real time -just by using the tm1 “wizard” approach and not wrting the IT Process Scripting?

Would like to know what are all general TI process we can create using the “wizard” approach and its advantages and disadvantages.

Regards,
Ravi

[ravi]

Corrction TI (TurboIntegrator) process

[lotsaram]

I do not believe there would be a single TM1 system in existence of any quality or utility for users that used only wizard generated code. The wizard and the script it produces is simply too basic and limited to do anything beyond loading data to cubes (and even then without any ability to handle errors or dynamically clear cube areas).

To build anything useful or of any complexity script has to be written.

[David Usherwood]

@lotsa, I agree - but I'm quite prepared to use the wizard when the requirement is simple enough.
Have you looked at 10.1 Performance Modeler yet?

[rkaif]

Personally I think using Wizard is only for the newbies who just want to very basic stuff. Wizards are also good for the Prototyping but for the 'real-world' projects you need to be good in TI coding.

[Harvey]

I occasionally use the wizard to generate code as a starting point for a TI process, then copy it out of the generated area and modify it (usually quite heavily!).

However, most TM1 developers will have built up a library of TI code over time, and will prefer to re-use what they can of that, often in combination with Bedrock.

[asutcliffe]

Like most others on this thread, I hate the wizard and almost never use it (with the exception of the occasional ad hoc process to move data around).

However, where I now work, many of the existing processes have been built this way. Users are comfortable with this approach and can create and fix things themselves. As much as I'd love to replace them all (no doubt reducing the total number of processes by about 90%), I'm finding it difficult to justify replacing something that works with something (at least perceived to be) more complicated.

So as hopelessly limiting as I find the wizard personally, I must admit that it has enabled people with almost no TM1 experience to build 90% of something they find useful.

[paulsimon]

I have recently worked at a client that tried this approach. They had little TM1 expertise but lots of Oracle expertise. They tried to do all the ETL in Oracle and then just use the Wizard to load the final results from the SQL View. However, this approach did not work and they had to resort to a lot of TI scripting. It might work on a small system, but I doubt that it can work on anything large or complex.

I can understand the argument about users being able to maintain a system even if it is less than efficient, in terms of speed or amount of time spent on coding. However, in my experience a system is generally developed by 1-3 people, and used by many more. Therefore 1-3 people need to be trained to use TI script, not every user. You can however reduce the complexity of some of the scripts to match the experience levels of the users. With the best will in the world if it is part of their job alongside eg maintaining the general ledger then they are not going to become as adept as a TM1 consultant.

Another approach is to provide a library of processes and a framework within which to work, that allows them to do complex things simply by calling pre-built processes. For example, a common issue is handling Orphans, ie elements without parents in the hierarchy. If you can have a framework or set of standards that says that each hierarchical dimension has a Top level consolidation and an Orphans consolidation feeding in to this, and that the Top is called All <dim Name> and the Orphans consolidation is called Orphans <dim name> and you provide them with a process called eg z_Gen_Handle_Orphans_Param_Dim. Then it is relatively simple to tell them to paste an ExecuteProcess('z_Gen_Handle_Orphans_Param_Dim','pDim',vDim) into the Epilog, and the process will then identify any Orphans in the dimension and link them to the Orphans consolidation, which in turn links to the top level. It is then relatively easy to build a spreadsheet that will show them any dimensions that have any elements under the Orphan consolidation.

Regards

Paul Simon

[jim wood]

I'm with Lazarus. I use the wizard to generate code that can save time, but then like everybody else I move away from the wizard quickly. I am however of the opinion that if the requirement is simple enough (simply clearing a cube) then leaving it within the wizard isn't that bad, especially if working for a company that has no previous TM1 experience.