Data Manager related - need help-thank you

[wprune]

Hi Cognos gurus,

My company bought a data manager as our ETL tools and still in the midst of learning stream behind it  ::)

There are some questions regarding DM =
1. How we define how many templates in 1 dimension should we have?
2. What is the function of lookup table? How to define how many lookup table should we have?
3. What is the relationship between dimension and fact build when we execute the jobstream? Need an explanation the process behind it..

Regards,
prune

[rockytopmark]

1. 1 Template/Dimension Build

2.
  a). Used by Dimension Builds to provide the structure and source
  b). Used in Fact builds to "Lookup" the Dimension's surrogate ID via the business key.
  c). How many is dependant on too many things to list.

3. Dimensions first, Facts after dimensions are done.

Have your team read books by Ralph Kimball to get a better understanding of how a Data Warehouse is designed, and built.  If the team doesn't understand the basic principles and proven practices, the endeavor will likey be less than successful.  Data Manager is just a tool that can assist in this, but it is good for following these aforementioned principles and practices.

Good Luck!

[MFGF]

Hi,

1. A dimension may have many templates.  When you define the hierarchy which describes the structure of each dimension, the hierarchy will need to use a minimum of one input template, but could use up to one input template per hierarchy level - the choice is yours.  When you define the dimension build to deliver your dimension data from the hierarchy, the dimension build will need to use a minimum of one output template, but could use up to one output template per table if you are delivering a snowflake schema - again the choice is yours.  It's also possible to re-use an input template as an output template, so the bare minimum you could get away with is one template per dimension, but you could potentially have many.

2. A lookup is a single-level dimensional structure which gets built and cached in memory.  Usually a lookup is used to allow you to validate your business keys coming through a fact build, and to replace the business keys with surrogate keys in the fact build.  Other common uses for lookups are as a translation mechanism - converting keys or strings or currencys or...  You would create as many lookups as required - usually you will need at least one per dimension for BK validation/SK substitution, but data translation requirements may result in you creating others too.

3. A Jobstream is simply a mechanism of allowing you to implement process control.  Typically you will end up with a large number of dimension and fact builds which then need to be run in a specific order with checks to make sure nothing is going wrong.  This is what a jobstream is used for - it allows you to define how the builds should be run, and it allows you to intersperse these with other logic where required.

Regards,

MF.