Thoughts on Business Intelligence Data Architecture
In our company we use the Zachman Diagram to complete all the architecture artifacts. On the other hand, we have SAP as main system for all our administrative processes. In other words, we have bought a packaged solution with a minimal insight in the physical data model.
What we do have is a logical data model for our company. Derived from that logical data model, we could abstract a logical Business Intelligence Data model. Since SAP BW is our core BI system and we use the Business content we have a preconfigured Business Intelligence data model delivered. Now we are in trouble! Our business people are getting used to the terms and language used in the logical data model. On the other hand they are confronted with the SAP minded terms and language in the Business content Infocubes and queries. So there is a gap between the business language and the SAP language. There are some people who want to go all the way to map the business content to the SAP tables in ECC. I do not agree to this, since this is very labour intensive. I am still wondering how we can cross this bridge of mis-interpretation.
Further complication: there are a lot of data model standards in the industry. We are in the process of adapting several standards to improve interoperability between our own internal systems (including SAP) and our outside world which may or may not use SAP. We use the Common Information model (CIM for utilities) as our reference model.
Somehow this all the data models, semantics and techniques have to be melted together. This seems to be a huge task, I hope this will not result in a melt down!
In my upcoming blog I will write about the large buidling blocks we have in our BI logical architecture. This will give you insight in the complexity and how we try to overcome this.