Looking into creating a Datawarehouse 1

[eo]

And I am an absolute novice, so please bear with me. We are looking into creating a datawarehouse, and an executive desicion was made to try and keep as much as possible in-house. This therefopre comes down to yours truely!!! I will therefor have to go on urgent cources to bring me up to speed on this, so that I can work with the consultants. What I am unclear of, is how does the whole OLAP and DW issue fit together? I am well aware of what OLAP is (data storage cubes) as we use them in our MI function. But is all DW'es OLAP cubes. Can anyone enligten me, or point me into a informative direction? I need to go on to relevant courses in this regard.

Many thanks,

EO
Hertfordshire, England

 

[blom0344]

To keep matters simple: Try to think in terms of backend/frontend.
The ods/datawarehouse/datamarts make up the backend and the frontend will consist of the reporting/analysis tools that you intent to use on the backend.

OLAP cubes is just ONE way of creating reporting structures based on the backend.

It would be quite unreasonable though to ask us to tell you what to do. This is a technical forum and we can advise you on technical issues. You can find all sorts of links if you search the forum though.

Keep in mind that there is no such thing as one winning strategy. If you succeed to satisfy the business over a period of time you can regard the implementation a succes.
Even if it deviates from all the text-book examples..

Ties Blom

 

[eo]

Thanks for this, and I merely asked this Q as I want to clarofy where to go from here, and will take much more into account.

Sticking to your BE/FE example. Just one more thing:

We have a range of SQL and Oracle Relational DB's. From this we aim to create appropriate OLAP cubes to easy the reporting and analysis process. Does this then imply a DW of sort has to exist as the OLAP is more the FE reporting structure. Or would the OLAP cube in this case be its own back-end.

I hope I am making sense.



EO
Hertfordshire, England

 

[blom0344]

No, the OLAP cube is basically a way to feed information to the user. Every cube will be build from a relational structure that exists in terms of data/dimensional tables.

OLAP cubes are potentially very powerful, capable of getting large volumes of data across (either online or as offline versions). However do not limit your front-end to cubes. If there is a demand for lots of ad-hoc analysis on the datawarehouse , then cubes are not always the best option.(less flexible) Sophisticated analysis tools as Business Objects provide much more freedom in that respect.

Most datawarehouse implementations will be the basis for both large volume fixed reports, as well as providing analysts proper sources for analysis.

If you have many different sources then consider 3-tier solution: ODS / datawarehouse / datamarts(cubes)

Ties Blom