Creating Production Dashboards

[Zimbo_African_1980]

Hi All,
I am new to dashboards in Cognos Analytics. My prior years have been in report studio.
Fortunately, I started a new job and am the team lead for Cognos Analytics but more on the dashboards feature.
I have already created some samples using flat files and so far soo good.

I would like to know how this application works from the framework Manager Package to the module/dataset?
What are some of the best practices.
Should I use a dataset or module as data source for my dashboard?
Our main database is refreshed every 15 minutes, so what would be my best option to build a dashboard that I can also refresh every 15 or 20 minutes.

Also looking for conditional formating and use of third party add-ons if possible.

If there is a thread that you can point me too, that would be awesome..

Thank in advance,

Arnold

[MFGF]

Hi All,
I am new to dashboards in Cognos Analytics. My prior years have been in report studio.
Fortunately, I started a new job and am the team lead for Cognos Analytics but more on the dashboards feature.
I have already created some samples using flat files and so far soo good.

I would like to know how this application works from the framework Manager Package to the module/dataset?
What are some of the best practices.
Should I use a dataset or module as data source for my dashboard?
Our main database is refreshed every 15 minutes, so what would be my best option to build a dashboard that I can also refresh every 15 or 20 minutes.

Also looking for conditional formating and use of third party add-ons if possible.

If there is a thread that you can point me too, that would be awesome..

Thank in advance,

Arnold

Hi Arnold,

Dashboards are a great way of visualizing your data quickly and easily, with automatic interactivity included between the widgets. There is a powerful viz recommender that tries to choose the most appropriate way to visualize the data you bring in, but you also have the flexibility of choosing specific visualizations and editing the linkages between widgets.

You can build dashboards directly over uploaded data sets, over heritage packages published from Framework Manager, over data modelled in Data Modules or over data extracted into data sets. The most robust choices are packages and data modules built over underlying databases - often the data is governed and secure, and as the databases are updated the data automatically flows through. If you choose uploaded files or data sets then things get more tricky in terms of seeing updated data. You'd need to re-upload any modified files, and schedule data sets to refresh on whatever timetable is required.

You can get the widgets in your dashboard to automatically refresh. Open the Properties pane, select a widget, and you'll see the option in Properties. You'll need to set this for each widget. I think the minimum interval is 5 seconds, but in your case you'd just set your widgets to refresh every 15 or 20 minutes - whatever you choose.

There is some simple conditional formatting available, but nothing as detailed or powerful as you'd find in Reporting. It may or may not be suitable in your case - it depends what you need to do.

Cheers!

MF.

[bus_pass_man]

Define the problem.   Don't look at it as a this tech vs that tech.

The primary consideration must be business requirements. What is needed? Where are there problems and unmet needs?

If you want to use uploaded files so that users can have some degree of self-service, what controls, training, and guidance are you going to provide? How are going to prevent someone from making decisions about things based on a dashboard which is wrong, but not obviously wrong?  How are you going to provide governance on the data being used by users in their self-service situations? You might want to look into using the SDK to create an app which automatically updates the uploaded files which your users want to use.  You could use that to formalize the process.  Keep in mind that you and your users might need to document what data you are using, its provenance, and what processes you used to ensure proper governance.  Some of the people to whom you might need to provide this documentation to include senior management, internal and external auditors, and the old Bill.

How do you prevent overconfidence causing reports and dashboards being created by users, which are incorrect?

What sorts of tradeoffs are you going to be facing?

What investment in legacy apps do you have? 

In general, you would want to leverage what you have by importing the FM packages into data modules and allowing users access to them in order to create relationships between objects in the FM package and their uploaded files.

Whether you want to create views based on things in the package or allow users to create relationships to the objects in the package is something you would need to consider.  The former would allow you to create objects with all the dimensions conformed to a particular fact table although you would need to ensure that your determinants still work and if not, create column dependency on the views.

There is a certain inescapable minimum amount of knowledge that people need -- concepts such as relationships, keys, cardinality, fact grains, dimensions versus facts -- in order to be able to self-serve themselves.  A key doesn't need to be an abstract number, although it helps for query performance.

Here's some best practice docs.

https://community.ibm.com/community/user/businessanalytics/blogs/torben-noer1/2020/09/08/metadata-modeling?CommunityKey=6b10df83-0b3c-4f92-8b1f-1fd80d0e7e58&tab=recentcommunityblogsdashboard

https://www.ibm.com/support/knowledgecenter/SSEP7J_11.1.0/com.ibm.swg.ba.cognos.mod_guidelines.doc/c_mod_guidelines.html

https://www.ibm.com/support/knowledgecenter/SSEP7J_11.1.0/com.ibm.swg.ba.cognos.ca_mdlg.doc/c_ca_data_modeling.html

https://community.ibm.com/accelerators/catalog/content/Metadata-Modeling-Best-Practices

https://www.ibm.com/support/knowledgecenter/SSEP7J_11.1.0/com.ibm.swg.ba.cognos.mod_guidelines.doc/mod_guidelines.pdf?view=kc