Differences between OLAP and Data warehousing concepts? 4

[bironman]

Newbie question:<br>
<br>
What's the differences between OLAP and Data warehousing concepts? Could exist Data warehousing without OLAP?<br>
<br>
By the way, why each concept have a different forum?<br>
<br>
TIA

 

[jnicho02]

Data warehousing is about combining data from a number of different sources in a company... e.g. from the sales system, the stock control system, sales forecasts, etc. The data can then be accessed within the same queries.<br>
OLAP is about querying data in a multi-dimensional way.<br>
<br>
A DW is ideal for feeding an OLAP tool as all the required data is held in one place and is usually structured in a way that suits the queries.<br>
<br>
Furthermore, an OLTP system usually holds the current picture (e.g. how many of Product X are in stock at this exact moment?) whereas a DW may hold historical data (e.g. how did a hot summer effect our stock levels?)<br>
<br>
[I await further refinement from WHArchitect]

 

[SASMAN]

Bironman,<br>
Good question,<br>
Short answer:<br>
The information delivery environment in most organisations involves extracting data out of the OLTP (operational) databases, manipulating and transforming this data a bit and then storing it along with data from many other disparate sources structured according to subject areas.. Essentially, this is done because its impossible to provide the kind of business reports needed, directly from the OLTP databases because A: you may impact the performance of those systems and B: they dont store history in them thar OLTP dbases. Collectively the activities associated with extracting transforming and loading data into a structure to support the information delivery requirements of an organisation is called data warehousing. Did someone mention metadata?<br>
<br>
OLAP on the other hand, deals with exploiting the data already stored in a data warehouse. Normally, from a technical point of view, an application that gives a user OLAP functionality involves a model and a viewer. The model handles access to data and the viewer handles display of information. The type of functionality provided includes multidimensional slicing and dicing of data (a hard concept to describe in words, but pretty easy to demo with the right viewer), drill down, expanding, rotating........you have got to see these things to appreciate what they are.<br>
<br>
"Could exist Data warehousing without OLAP?" Yes, there are plenty of ways to exploit data warehouses besides OLAP. Think of data mining for example or a customised business solution such as risk management that uses OLAP as just one of many tools to achieve its goals all of which are down-stream, dependent on data warehousing.<br>
Cheers<br>
SASMAN

 

[WHArchitect]

There is a simple answer to bironman's question and jnicho02 hit it pretty much on the head. Basically, a data warehouse is a database containing data integrated from a number of data sources. OLAP is one way to read the data.<br>
<br>
There are many advocates of "three tier architecture" for a data warehouse, in which application systems contribute data to a central warehouse which is used to populate data marts. I am among them. However, a central warehouse is not an absolute necessity if one wishes to deploy an OLAP tool. as jnicho02 pointed out, a warehouse is an ideal source of data to populate a multidimensional database or a stand-alone star schema. It is not the only possible source, however. For instance, most ERP vendors bundle an OLAP tool with their software. This is not to imply that the integrated database that is the salient feature of this kind of application makes an adequate substitute for a data warehouse. This is far from true. ERP providers include the OLAP tool because data warehousing in general and OLAP in particular are hot right now and the tool can actually be used to some advantage in the ERP environment. OLAP tool providers enter into these agreements because it permits them to inflate their market share figures and imply a level of customer satisfaction that is less than accurate.<br>
<br>
The whole point is that OLAP is a way of looking at data. It does not require a data warehouse to function properly. In fact, many IT organizations have rolled out multidimensional data marts without the benefit of a central data warehouse.