Author Topic: Transformer Cubes  (Read 524 times)

Offline cognos3

  • Full Member
  • ***
  • Join Date: Jun 2012
  • Posts: 9
  • Forum Citizenship: +0/-0
Transformer Cubes
« on: 01 Dec 2017 12:16:15 pm »
We are having issues with our Dashboard reports performance which are based on FM packages and we are thinking to start using Transformer to build some cubes(company bought Cognos products in 2010 and haven't used Transformer yet :) ).

Gurus- Can you please provide your valuable input if it's a good idea to build a cube to enhance performance? or Should we go through some other route (may be using qlikview)?


Offline bdbits

  • Super Moderator
  • Statesman
  • ******
  • Join Date: Feb 2010
  • Posts: 1,799
  • Forum Citizenship: +105/-0
Re: Transformer Cubes
« Reply #1 on: 04 Dec 2017 04:26:27 pm »
Transformer is a legacy Cognos product. If IBM could drop it without angering customers, they probably would. I would not expect any future enhancements. As such, I would not start using it without a very good reason. But all is not lost - there are contemporary cube options in Cognos like Dynamic cubes.

Before going down that route, I would consider your database design, and the impact of the queries being generated.
* Is this a data warehouse, and if so do you have star schemas?
* Have appropriate indexes been created?
* Have you considered other database options, this will vary with database of course but things like materialized views, columnstore/in-memory options, etc.
* Does the database server have adequate resources - RAM, disk space (especially temp storage), CPUs, etc.
* If you run the query being generated by Cognos, how quickly is the result set returned? (It may not be the database, so a cube may not help.)
* If all of the above have been considered, do the types of queries lend themselves to cubes? I would argue you mostly want aggregate queries to take advantage of cubes, as opposed to transaction/fact-level result sets.

Hope that helps.