Performance

[José Luiz]

Hello,
I have some queries that are JOIN with 25 tables in the database, right? but the performance of the system falls far. What I did then, generated in a single table all records but with 278,000 lines and performance is also bad, to get an idea each query takes 5 to 6 minutes ..

Is there a procedure that can do?

[AlexZencovich]

Sorry, may be I miss some - each query from 25 table takes 5 or 6 min? Which DB source you have used?

[José Luiz]

Hi

i use sqlserver 2000 but i have one component query (Tquery) and comand is

select * from dwClients ( where dwClients is table contain 278,000 lines)

You have solution, because my result between 5 and 6 minutes

Best regards

[AlexZencovich]

how many dimension and measures you have?

btw - check if you do not get RecordCount property somewhere. If you read that property - avoid it.

[José Luiz]

I have 13 dimension and 5 measures.

btw? whats it is ?

I'm not using any command to count the record, but it can be an active component of the property?

It is correct I only have a table with all the records?
or the right is to have a separate table for example, City, Customers, Products, Sales?

I'm sure I'm doing something wrong, because it takes too much consultation, I see that the demo of the component does not take much, but I can not find where it is.

Best regards

[AlexZencovich]

I have 13 dimension and 5 measures.

btw? whats it is ?

I'm not using any command to count the record, but it can be an active component of the property?

It is correct I only have a table with all the records?
or the right is to have a separate table for example, City, Customers, Products, Sales?

I'm sure I'm doing something wrong, because it takes too much consultation, I see that the demo of the component does not take much, but I can not find where it is.

Best regards

1. BTW - By the way

2. Yes, better to have separate tables/quesries for dimension items and indices to items in fact table - it allows load dimensions BEFORE fill facts (Star schema, see PivotCube manual). SOmetims it is impossible (for date field for example) but anyway change Table schema to Star wil increase perfomance

3. I suppose you have probelm with with cache in somewhere in SQL driver. Try to do next

1. Do not open query/table before start build cube
2. Make example code like
i := 0;
Query.Open; // facts query
while not Query.Eof do
begin
Query.Next;
inc(i);
end;
Query.Close;
ShowMessage(IntToStr(i)); // check if actual value like you expect

[José Luiz]

Hello,

btw hehehe

Okay, I'll separate the tables and then relational.

I hope the performance increase ...

best regards