MDM (Master Data Management) has been a buzz word past several years. Most of the Information/Data Architects are beginning to understand how to design and implement MDM and use it as an SOA enabler.
There is another concept called Information as a Service (IaaS) that goes along with MDM. By definition, IaaS is an intermediary layer between your enterprise data repositories and common SOA service layer.
My question though is about implementation of IaaS. At the beginning, I thought that DB Stored procedures are good enough to establish IaaS layer. But since your data is coming from multiple disparate database platforms (Oracle, SQL Server etc) and geographical locations, stored procedures cannot be use开发者_JAVA百科d to form consistent IaaS layer.
I've done a little research that leads me to several ideas. I would like to get some more insight, tips and tircks about IaaS implementation from you fine data professionals who already established IaaS in their environments.
It's not exactly what you asked for, but there are 'Federated' databases that gives you a standard DB abstraction to data sources that are spread across different physical databases, from different vendors, and some of the support also other data sources (non databases).
I know specifically of the IBM product (which is part of the DB2 suite) but other vendors have similar tools.
You should note however that having both a data abstraction layer, and a business abstraction layer (Business Services) may turn out to be an overkill - depending on the specific scenario.
精彩评论