by Rosario E
Data warehousing has been evolving since 2008 to analytical architecture that includes data marts, ETL, near line storage, exploration warehouses.
The SQL Server has also evolved since started with small amounts of data on a personal computer with basic functions until now that serves mid-size and large amounts of data for warehousing and it is one of the preferred technology platform for advanced data warehouse.
Before the preferred storage medium was disk storage. This also allowed online transactions, with the creation of indexes, which required a rapid and random access of data.
So for certain online applications using a rapid, random access of small amount of data is fine, and when it comes to an analytical process it is different, since is done by SQL which are access sequential.
Another problem could be the way the organizations build their data marts before they have build a data warehouse. the migration of data becomes difficult when you have to go from multiple data marts to a data warehouse, and some times this data marts needs to be rewritten.
Also the cost of data warehouse it could no be and issue today, but when the amount of data start to increase, also the cost start to increase and this will depend on the central processing units a organization needs.
Also the ways to manage data is another aspect of cost, if it needs to be compress it will not update, because to be update it needs to be decompress to update and compress again, and is costly.
One of the gain it is parallel processing which means data could be stored on more than one device, so more than one processor could access and manage data a the same time. so this means data could be divided in more than one server. in other to have more efficient access to data most of the high performance data needs to be separate from the slow performance so it needs to have an efficient way to separate one from the other.
Also a large data amount need to be capture in a rapid and predictable way triggered by a unique event it is called streaming mode. there are a different in static data and streaming data, and active data but it does not need to be treated as statistic data that requires an infrastructure the requires integrity so each of them should be treated as they are.
W.H. Inmon. Published October 2009, SQL Server 2008 R2, Data warehousing 2.0 and SQL Server: Architecture and Vision retrieved 29Jan2014, from http://msdn.microsoft.com/en-us/library/ee730351(d=pinter).aspx