Sunday, February 2, 2014

A Master Data Mind Map | LinkedIn Group: DAMA International

Follow the LinkedIn discussion

My comment 

Starting with a mind map is definitely a more "relaxed" technique during the brainstorming stage of an MDM endeavor, compared to immediately using a data modeling tool. The relationships between master data entities are simply specializations, so, at the first stage, the creative process is not overloaded with the pressure to name relationships and assign cardinalities.

Therefore, this approach is more suitable to obtain acceptance from the target audience (stakeholders / representatives of the business units). Their engagement and contributions are not only indispensable to find and define master data entities as the center of operational transactions, but also to build a sustainable basis for analytics processes: As you mentioned in your blog, master data are about the Who (Party), What (Product / Service) and Where (Location), i.e. these master data entity categories (together with When = Time) also define the dimensions of the analytics space, as most business questions can be projected into and then answered from this 4-dimensional structure.

Modeling of un-structured data | LinkedIn Group: Data Architect USA

Follow the LinkedIn discussion

My comment 

All business-relevant data should be modeled in only one tool to ensure the integrity between logical enterprise models, logical subject area models and application-specific DBMS models.

Differences between modeling techniques for SQL databases, NoSQL databases and other storage / retrieval technologies only occur on the physical level. In a nutshell, while SQL database models 'may' be denormalized, NoSQL database models 'must' (almost always) be denormalized (due to rare to no support of table joins).

In any event, professional data modeling tools (should) offer such denormalization support while keeping track of the column lineage.

The Business Architecture tool SILVERRUN in its latest version (published last week) features modeling Cassandra 2.0 databases incl. the generation of CQL scripts (see http://www.silverrun.com/silverrun-news-nosql-cassandra-modeling.html ) Additional SILVERRUN versions including reverse engineering of Cassandra databases as well as support of other NoSQL databases will follow in 2014.

[In the spirit of full disclosure: I am in charge of Grandite, the SILVERRUN supplier, and will be happy to provide additional information on demand.]