Roelant Vos An expert view on Agile Data Warehousing

Enterprise Architecture seminar thoughts 0

Enterprise Architecture seminar thoughts

 

 This posting is about the most complex object that humankind can come up with; an Enterprise. At least that is what John Zachman states at his one day seminar about Enterprise Architecture (physics 101 – Monday 23rd of August 2010 – Sydney). The first rule for the way of working is defined right at the start: when doing anything with an enterprise or start designing anything for an enterprise; make sure you understand the enterprise...

Completed CBIP certification 3

Completed CBIP certification

 

 As of July 2010, Roelant Vos is a certified business intelligence professional (CBIP), specialized in data management! The CBIP exam involves a minimum of three examinations ranging from generic IT to datawarehousing and data modelling. I’m very happy that my employer Analytics8 made this possible.

Why you need an Enterprise BI framework 0

Why you need an Enterprise BI framework

 

 Over the last few months I have done a number of pre-sales presentations regarding a framework for designing and developing data integration programs. While the framework in question is a smaller scale (mainly ETL) I did encounter numerous discussion why such a framework is necessary in the first place. What would be achieved by using it? Because this blog serves as my own personal framework and collection of best practices I thought it would be...

Current mapping generation improvement points 0

Current mapping generation improvement points

 

 I’ve demo-ed the posted processes for mapping generation and Data Vault a couple of times for varying audiences and (lucky me) no one has noticed the existing flaws 🙂 So it’s time to make a list of them to keep me from not forgetting to fix them. And it’s a heads up for the few people I know who are using the scripts. There is currently no way to keep the ‘transactional’ attributes out of...

Mapping generation for Data Vault demo: part 6 (links) 6

Mapping generation for Data Vault demo: part 6 (links)

 

 Getting the Data Vault link entities right is probably the hardest part of the mapping generation algorithm. The following script will require user input at critical moments in the process. It helps to have the target datamodel at hand so you know why (and how) these choices are made. By reading from the history area tables and using the Hub entities created in the previous step this algorithm will create relationship tables (links) including link satellites....

Mapping generation for Data Vault demo: part 5 (hubs and satellites) 0

Mapping generation for Data Vault demo: part 5 (hubs and satellites)

 

 After the successful creation of a history layer, it is time to focus on the core parts of the datawarehouse: the Data Vault. In this example a ‘raw’ Data Vault is created: no changes or transformations are done here. The first script in this layer creates the Hubs and the corresponding Satellite. Currently this is done based on the history layer, which is probably not the best approach. It would be better to have the...

Mapping generation for Data Vault demo: part 4 (history area) 0

Mapping generation for Data Vault demo: part 4 (history area)

 

 Now that the staging area tables and the mappings from source to staging are created, it’s time for the next step: the history area. In this step the source data is archived in a SCD2 way. The source tcl file for the staging to history mappings can be downloaded here: 2_staging_to_history_generation. When run (source command!) the script will ask the familiar questions about replacing ETL metadata. The script executes the following steps, specific elements are...