For many years I have been advocating data warehousing and BI. Large enterprises like global banks need to integrate data from many systems and my answer has always been building an Enterprise Data Warehouse. It takes years to build data integration and EDW. And often it covers only a few departments, not the whole enterprise.
Data virtualisation on the other hand, enables us to integrate data from various transaction systems/applications, databases, DW, files, spreadsheets, the Cloud and Big Data within 3-6 months, making us more agile and cost effective.
While data virtualisation doesn’t provide the solid data platform that data warehouse provides, the speed and cost of building it provides huge business benefit. In a data warehouse the data is very well arranged and very user oriented. In data virtualisation we will still have all the data structure issues from the source systems. This is the main drawback.
Above is how I see it in large corporations. It takes 12-18 months to put together an enterprise data warehouse (first go live 12 months) whereas had we used data virtualisation it would only take 6-9 months (first go live 6 months). To build EDW we need 10 people, whereas DV we only need 5. To build EDW we need design tools, ETL tools and database servers such as ERWin, Informatica and Oracle 11g, whereas DV is more light weight tools such as Composite.
Data virtualisation will not be able to replace a data warehouse. But it will complement it. It is a quick tactical win. But usually, a tactical initiative becomes permanent solution.