Cognos Analytics with Watson

 View Only

Starting a new modelling project - FM or data module

By Michael McGeein posted Wed May 15, 2019 01:50 PM


30 July 2020 - Please refer to Framework Manager features not supported by data modules for a more up to date list. You can always refer to the list in the knowledge center at a later point as well because we will be updating it as we address the gaps. 

As a modeller or business analyst starting a new project, you may wonder which capability to use to create the model. In Cognos Analytics there are two primary methods for building a model: . If the modeller is familiar with one of these capabilities, (s)he might start, as I would, with ‘what I know.’ In your career you likely have had to change ‘tools’ for any number of reasons: tool no longer supported, employer changed products/tools, existing product does not support the task you must have access to, etc. You don’t need to do that in CA. Use the modelling tool you are most familiar with.

When you are building a model, the focus should be on the model, not the modelling tool. There are some features in each of FM and DM that REQUIRE you use THAT tool. For example, as of 11.1.2, Object level security is only supported in FM. If you need that, you NEED to use FM. Relative dates were added in 11.1.x, but only in DM. If your users need relative dates, DM is your only choice.

We are continuing to evolve DM, to include many of the capabilities for FM within DM. However, there remains a small feature parity gap. Listed below is a comparison grid of FM vs DM for each of 11.0.x and 11.1.2.  Please review it carefully, prior to starting a new project, to avoid creating duplicate work for yourself. Choosing incorrectly and starting in one and finding out you need the other to address the business/modelling requirements identified by your users can be frustrating.

As a general rule, I suggest to new users who are building models, start with a data module, given its simpler, more intuitive web-based interface, targeted at business users.FM-DM 11.0 vs 11.1




Thu July 30, 2020 05:44 PM

Just one thing.
In Europe, the medium and large companies report in several languages. I recently started a new project, a small project, but in French, English and German. I am not going to maintain three projects, one per language!

In fact, anything you do in DMon can be done with FM and a lot more.
I suspect that:
1. The people who are currently developing do not really know FM,
2. That the developers mainly want to have fun and the needs of the users do not really interest them.
Two weeks ago one of my clients - and this is not the first - decided to keep the existing Cognos reports - they have a lot of them - but all new projects will be done with another tool.
I'm starting to be convinced that Cognos BI will end up like Lotus. :(

Mon August 19, 2019 04:42 PM

It is possible to easily create simple models with FM. In addition it is certain to be able to evolve them.
It is true that a simple tool for simple cases is useful, it is not possible to say the opposite. But in this case why reproduce features that exist in FM and are intended for "entreprise models". As for example multilingualism.
If I understand correctly IBM wants to give up FM - no evolution means for me abandon - and it's not a good idea. I work with a dozen companies and the DWH remains the most used data source for BI (and not Excel files). It requires an adequate tool.
Try to recreate "Go data Warehouse" model with DM, it is not a gigantic model. I do not think the result is convincing.

Mon August 12, 2019 03:39 PM

Since Data Modules are lacking the parameters and some other functions, has anyone tried applying the security in framework manager and then linked the package with the security to table in a data module?

Thu July 25, 2019 01:51 PM

It's true that Framework Manager is a better tool than data modules for modeling out an entire kimball warehouse or a 350 table operational system.

It's likewise true that nobody wants to wait three months for you or I to build that gigantic FM model anymore, especially when what they need can be found in just three of the 350 tables.

Data modules are far superior when it comes to enabling rapid turnaround of small to medium sized models tailored to answer a discreet set of business questions. Whether or not you or I think this is an appropriate way to model data is irrelevant, as clearly demonstrated by the gargantuan shifts in the BI market over the last 5 years.

Thu June 06, 2019 12:07 PM

I agree wholeheartedly with the two comments. DM is fine for the light-weight modelling tasks, but for the heavy lifting stuff it just doesn't provide the platform. DM seems to be a tool for report authors who want to pretend to be data modelers, and IBM is giving them a chance to play out their fantasies - kind of like Fantasy Football... *^_^*

Thu May 30, 2019 04:32 AM

Completely agree with this comment
And that's not all !
All the possibilities to drive generated sql code in data governors or query options.
DM is too short for a developper who works in an mid-sized to large entreprise. And with all this futures it is too complicated for "self service BI". CA 11 becomes a gas plant and it is imperative to structure it !

Sun May 19, 2019 05:29 AM

I think there is more to consider, e.g.
  - branching/merging/segmentation of models (FM only)
  - bulk search/replace (extremely useful - FM only)
  - durable models (need multilingual properties - FM only)
  - handling of large models (slow in Data Modules depending on workstation/browser)
  - persistent graphical representation (FM only?)
  - can generate documentation (FM only)
  - creation of DMR models (FM only)
  - testing objects (greater flexibility in FM)
  - transaction log/scripting/action replay (FM only)
  - prompt UI settings for adhoc Reporting (FM only)
  - no select and search prompt for data modules (

In my opinion, Data Modules won't be a choice for mid-sized to large enterprise models in the near future.

Sadly, new features aren't added to FM, so once again you have to make a decision with many prerequisites. This is the same when deciding which Dashboard technology to use (Dashboard/Legacy Workspace/Legacy Portal Tabs) or which chart engine to use (Rave2/Rave/Version 10/Legacy). All technologies have ther unique feature set.