Re: CREATE DATABASE using OCI, is it possible?

From: Job Miller <jobmiller_at_yahoo.com>
Date: 14 Mar 2002 14:49:00 -0800
Message-ID: <2edf8ed9.0203141448.3f7137ab_at_posting.google.com>


add a field to each of the four tables that stores the name. Then you will only ever have the four tables. instead of a different database for every new name that comes along.

you can create views based on the name column so that users can only see records from the tables with a certain name.

the only advantage to having this separated into different tables, is that you get smaller tables, but potentially hundreds or thousands of them.

this seems very simple, maybe i am missing what you are trying to do.

Job

rizzo2k_at_hotmail.com (rizzo2k_at_hotmail.com) wrote in message news:<6de30668.0203131255.301e8b3a_at_posting.google.com>...
> > Last but not least, I still don't really understand what you are trying to
> > do, so some (or all) of this may be very, very wrong advice indeed.
>
> Hi Mark,
>
> Thanks for your help. We are looking into using "CREATE SCHEMA" and
> some other things, but here is a more in depth explanation of what we
> are doing. (Short answer is that something has been dumped in our
> laps and we are trying to make the best of it.)
>
> We are basically working with a search and retrieval API that uses an
> SQL storage mechanism. The person who developed this developed it
> against mySQL. Currently, each collection of information is stored in
> 4 or so tables, lets say index, words, fields, and blah. Typically,
> this collection might be populated by a web crawl, ie. pulling down
> all documents linked from a certain web page. There is an index step
> that stuffs this textual data into the SQL tables appropiately. When
> this process is started a database is specified. If this database
> does not exist, then he would create it. This is what was meant by
> "creating a database on the fly".
>
> Each collection of information will have a name. In the current
> mySQL-based implementation, there was a database associated to each
> name. Each database/collection would then consist of the same 4
> tables just populated with different data.
>
> Based on what you and others have said, it looks like it would be
> best if we modify the existing search and retrieval API so that the
> table names are prefixed by the "collection" name. There will be a
> large number of these collections (4 tables for each collection).
> Moreover, there will be multiple users of the end application (There
> is just one "user" performing database operations, ie one oracle or
> mysql user). At some point in the future, individual users will be
> able to filter off certain documents from various collections to form
> their own collection.
>
> We expect that there will typically be between 100 and 2000 of these
> collections. The amount of text data being placed in these
> collections will vary from a couple megs to almost a gig. Once
> individual users are allowed to filter their own collections, we
> expect the number of collections to grow substantially.
>
> Any insight you have is appreciated.
>
> Thanks
Received on Thu Mar 14 2002 - 23:49:00 CET

Original text of this message