Need Some Information Reg. Data Warehousing???

[mallareddy8]

Hello Friends,

Please, I need some information about star schemas? i read the book data warehouse tool kit, i know what is star shema, what is fact & Dimension tables. what i need is
1. In RT(Real Time) how can we anlyze the data to create star schema.
2. In RT max. of howmany star schemas we can create in one DB.
3. In RT how big is each star schema?(means howmany fact tables? and howmany surrogate keys we can use?)
4. in RT shall we design more than one fact table in each star schema?
5. in RT shall we create seperate DB for each star shema?

thank you.

 

[srinathmk]

Hi mallareddy,

Data warehousing is about managing intelligently hystorical data. Hence,my assumption of the term RT(Real Time) is "Practical Situation". In this context the replies to your questions are as underneath:

1. Analysis of data is based on the following factors: Broad seggrigation of fields into Dimensions and Measures. This is done using user requirements. Also necessity of aggregate tables and look up tables if required are gathered. These are mapped to the source OLTP systems.The schema (Star/Snow flake) is decided as per practical needs.

2.From data warehousing front, there is no limitation on number of schemas on data base. This needs to be decided with the data base adminstrators on the feasibility and optimisation front.

3.Usually data warehouse schemas handle hystoric data. Hence the Star Schema. The data amount is not a constraint until the database supports it.

4.More than one fact table is possible. The Joins and looping needs to be taken care. This can be handled using aliases.

5.Creating a seperate data base is possible if you want it like that. This may serve for security purpose.

Regards,
Srinath M.K

 

[blom0344]

Real Time datawarehousing has nothing to do with Practical Situation (what ever that means) It is a new type or hype to get every mutation in the OLTP environment as quickly as possible into the datawarehouse environment (possibly with an ODS (operational data store) in between. This calls for new kind of ETL tools that rather PUSH data from OLTP to DWH than extract the other way. I have called it a hype, since the traditional type of DWH is still far from a succes and most DWH are supposed to support strategic decisions , which makes the availability of every last mutations a bit superfluous. A real satisfactory DWH in the traditional sense is quite something to achieve. Invest only in REAL-TIME stuff if the benefits are there.....(and proved) T. Blom
Information analyst
tbl@shimano-eu.com