Re: Uploading data for star schema

Hi,

This won't work for several reasons. For clarification: 1) I will be uploading several million rows and Access would explode 2) This process will be done regularly and some of the data will already be in the database so if the record is already there I need to get the id. If the records isn't there I need to pull the next valid id from a sequence in Oracle.
3) That process is not automated (realized I didn't specify this) and this will be done weekly in an automated fashion 4) This process will be far slower performance wise than what I am already doing with VB.

j_beverly wrote:

> On Tue, 05 Dec 2000 19:53:55 GMT, sw_at_weinerfamily.org wrote:
>
> >I have a textfile that contains rows of data in a non-normalized fashion
> >
> >What I need to do is load that data into a star schema.
> >In order to this I need to load each of the dimension tables and get the
> >ids of the entry. Of course if the row already exists I need to just get
> >the ids. After I have done this on all measure (8 of them). I need to
> >stuff the ids and other data into a record in a fact table.
> >
> >Currently, I am doing this using a VB program and it is incredibly slow.
> >Is there a way to this using SQLLoader or some other technque that would
> >be more efficient?
> >
> >NOTE - probably doesn't matter but I am not using Oracle OLAP.
> >
> >Thank you.
> >
> First of all, I'm not sure I complely understand what you're trying to
> do.
>
> You say your text data file is not normalized. I had a similar
> problem.
>
> In this case the legacy data came another (PROGRESS) database program
> that was poorly designed (not normalized) and on a different operating
> system. I used Microsoft Access as an interim step to massage the
> data to match the format of my Oracle database. (I am more used to
> using Access than Oracle).
> Then I ran a query/report in Access that generated a text file that
> matched my Oracle database, so the SQL Loader process was simple and
> ran very fast. (Okay, after many attempts to clean the data and get
> the bugs out.)
> This was a relatively small database (< 20,000 records). I don't
> know if this procedure would work for you.
Received on Tue Dec 05 2000 - 23:14:02 CST