Looking for some best practice guidelines on how to model your metadata in Cognos Analytics Data Module? Then read on....
Metadata modeling has been a large part of using Cognos, all the way back from ReportNet through Cognos BI 8 and 10 and it's still fully integrated into Cognos Analytics 11. For some users, the metadata modeling part has been very simple and quick to handle where it for other users has been a longer project with 1000's of objects, folders, join etc. Framework Manager could handle it all but it also meant it was maybe a bit too advanced for users that only wanted to grab a couple of files and report on them. It's also a Windows installed program which is not what we want in a modern architecture with different OS's and the need to be able to handle metadata from a browser.
So, the purpose of Data Modules in Cognos Analytics 11 has been to provide a new approach on metadata modeling... making it simpler for first time / casual users to use and automate as much as possible with AI functionality and at the same time provide all the functionality an experience user with a complex setup need. Self-service is one of the big focus points with Cognos Analytics and for metadata modeling, we need more users to be able to create something by themselves. We have the govern enterprise approved modules and then we have end users that can enhance the govern modules with their own metadata. Everyone is using the same Data Modules and combine the information they need like Lego bricks. Govern, secured enterprise metadata with end user self-service! Most of the functionality from Framework Manager is now available in Data Modules and the remaining areas will be enabled in Data Modules shortly. For a full list of differences between the 2 tools are here: https://www.ibm.com/support/knowledgecenter/en/SSEP7J_11.1.0/com.ibm.swg.ba.cognos.ca_mdlg.doc/c_bp_dm_diffs.html
The metadata modeling process in Data Modules are a bit different from Framework Manager - one good example is the "Determinants" in Framework Manager compared to the "column dependencies" in Data Module - same purpose of using them, but we improved the process and made it easier to use and understand in Data Modules. We have an extensive documentation on the Data Modules here: https://www.ibm.com/support/knowledgecenter/SSEP7J_11.1.0/com.ibm.swg.ba.cognos.ca_mdlg.doc/c_ca_data_modeling.html
and we have several Business Partners that have created blogs with tips&tricks on how to use them - just search for "Data Modules Cognos Analytics". We also have a lot of best practice collected for Framework Manager and a lot of it is still valid for Data Modules. Modeling a star schema hasn't changed, so the classic metadata modeling guidelines are still relevant here.
But there was a need to rewrite the guidelines with Data Modules as the basis and this is what we announced for the Cognos Analytics 11.1.7 release and it's available now!Web: https://www.ibm.com/support/knowledgecenter/SSEP7J_11.1.0/com.ibm.swg.ba.cognos.mod_guidelines.doc/c_mod_guidelines.htmlPDF: https://www.ibm.com/support/knowledgecenter/SSEP7J_11.1.0/com.ibm.swg.ba.cognos.mod_guidelines.doc/mod_guidelines.pdf?view=kc
Metadata Guidelines are not describing step-by-step instructions on how to use Data Modules, but are giving the best practice advice's on how to model your data and get the best performance out of it. This is the first release of the guidelines for Data Modules, so expect updates to the document.... and we welcome your feedback on what's missing, could be improved etc - use the comment field below.
The Guidelines follow the workflow of model your metadata of gathering requirements, finding data, remove ambiguity, design, enhance, focus on performance and iterate.
Removing ambiguity has always been important to ensure end users understand the model, but with Cognos Analytics it's now even more important since the AI functionality also use that model to understand your data.
We have historically been creating very large Framework Manager packages, since it was possible and was working - but this also meant we ended up with some extremely complex packages, sometimes over 30-50 mb of metadata, that were tricky to change. The focus of Data Modules has been to create smaller, more flexible modules and then combine them. Another new option with Cognos Analytics is also the usage of Data Sets which is also described in the Guidelines along with Cache, Joins, Aggregations, Multi-fact, Multi-grain and a lot of other topics...
I hope you all take a look at the Guidelines and let us know what you think!
Happy modeling ;-)
#home #BestPractices #bestpractices #DataSet #modeling #businessmodeling #performance #datamodule