Groups | Blog | Home
all groups > sql server data warehouse > july 2005 >

sql server data warehouse : Which my star schema models is better?



Resant
7/4/2005 7:07:14 PM
I work in consumer goods company and now design OLAP Data Warehouse.
My OLAP tables consist of :
1. Customer - all our customer data
{CustomerID, CustomerCode}
2. Type - type of the goods
{TypeID, TypeCode}
3. Grade - quality grade of material
{GradeID, CustomerCode, TypeCode, Grade}
4. Receiving - material receiving
{ReceivingID, TruckNo, Transporter,...}

I've design 2 kind of Receiving_Fact and don't know which one is better
:
[quoted text, click to view]
- Receiving_Fact
{ReceivingID, GradeID, Measures}
- Star Schema
CustomerDim & TypeDim dimension will be related to GradeDim dmension

[quoted text, click to view]
- Receiving_Fact
{ReceivingID, CustomerID, TypeID, GradeID, Measures}
- Star Schema
CustomerDim & TypeDim dimension will be related to GradeID field in

Receiving_Fact

I hope my illustration could be understand well. Model 1 is like
snowflake where dimension related to other dimension. Model 2 is star
schma form which all dimensions related to fact table.

Please help me, which model is better? I'm really confused.

Thanks a lot
Jéjé
7/4/2005 10:17:57 PM
The model 2 provide a better performance both for querying and cube process
if you have 4 dimensions created (Customer, Type, Grade, Receiving) and if
your cube is optimized, then the process time is better (not link made to
process the cube).

The model 1 is slower, but reduce the disk usage in your database.
In this case AS will always do a link between the fact table and the grade
table to process the cube. (slower)

both schemas works fine.


[quoted text, click to view]

Resant
7/6/2005 7:08:41 PM
Thanks a lot, your answer is really help
AddThis Social Bookmark Button