Re: On the subject of Data Warehouses, Data Cubes & OLAP....

From: DataMan <dataman_at_ev1.net>
Date: Wed, 22 Oct 2003 20:23:14 -0000
Message-ID: <vpdppi7ccdn0f1_at_corp.supernews.com>


"Bob Badour" <bbadour_at_golden.net> wrote:
>"DataMan" <dataman_at_ev1.net> wrote in message
>news:vpdad0mimma52e_at_corp.supernews.com...
>>
>> Most OLTP systems do support the information requested. It's just very
 difficult
>> to retrieve.
>
>That is a remarkable and very interesting assertion. What can you offer
 to
>support such a remarkable assertion?
>
The support information or difficult to retrieve statement?

>
>> Additionally when the business requires enterprise level reporting
>> across the various functional applications, a warehouse becomes a
 necessity.
>> You know the scene, large company with 10 different order entry systems
>> for each functional area of the company (i.e. cost center). One for small
>> biz, one for consumer, one for z product, one for y product, etc. And
 of
>> course there was never any upfront effort by a data management
 organization
>> to ensure consistent representation of enterprise level entities. So
 you
>> end up with a hodge podge set of data that must be abstracted to make
 consistent
>> within a DW environment. And since you're already there go ahead and
 put
>> it in a dimensional format so somebody can understand it and easily pull
>> data/reports.
>
>What exactly is a dimensional format? How does it differ from >relational
representation?
They are both relational. Assuming both the application database and the data warehouse are both stored in relational databases. Was your question directed to differences between a data warehouse database (let's say OLAP for convenience) and an application database (let's say OLTP for convenience)?  The primary difference is purpose. OLTP must support consistent real time data. OLAP only need support historical reporting.
>
>How or why can it improve understanding? How have you measured >this improved
understanding?
I didn't say that it does. OLTP is better for understanding the business rules of an enterprise. A dimensional model only shows you what the business wants to report.
>
>What features does it provide that facilitate reporting?
>

 The data structure allows for easier retrieval. Received on Wed Oct 22 2003 - 22:23:14 CEST

Original text of this message