- What factors lead companies to select a particular architecture and
- How successful are the various architectures?
So below i've pointed the most interesting parts of this document:
- 39% of the respondents uses the hub and spoke model (Inmon).
- 26 % of the respondents have built and maintains the busarchitecture (Kimball).
- The hub and spokearchitecture requires the most time to roll out with an average of 11 months.
- The bus architecture takes about 9 months to develop and deploy.
- The hub and spoke and the busarchitecture are equally succesful.
- One third of the company's switched the architecture.
- 30% procent of the switchers switched from the hub and spoke to the bus architecture (AND back!). This is awkward! The supporters of the different architectures talked about the failures of the other(!). Perhaps the made initially the wrong choice for a architecture, based on the wrong assumptions?
- The hub and spoke model has higher technical issues and needed more expert influence.
- There is not a big difference between the hub and spoke model and the centralized datawarehouse.
- When the need for a datawarehouse is high (and you want quick results).
- High need of information flow between organizational units. I don't understand this. Perhaps it 's meant as the unstructured version of the dataintegration of the hub and spoke architecture?
- You want to build a silod system/ departmental datawarehouse (from my earlier post).
- When the information requests don't change that much (from my earlier post).
- The view of the ware house prior to implementation is more limited in scope.
- When there is a high need of integration between organizational units.
- The datawarehouse is viewed as strategic.
- Percieved ability of the in-house staff is high.
- When the information needs are not quite clear (from my earlier post).
- When the information request are varied (from my earlier post).
- when information requests are very ad hoc (from my earlier post).
- The 'system of record' principle is important (from my earlier post).
- When there is (more) time to build the datawarehouse.
Hennie
Geen opmerkingen:
Een reactie posten