At 06:18 PM 4/6/00 -0000, jerzy.tomasik@... wrote:

>
>
>Some of the portability items that are important to me are:
>
>1. Oracle functions. I think that a UDF library patterned on
>functions available in Oracle would be very useful.
>

Probably doable, though there are probably name space conflicts
that would need to be handled somehow.

>2. Database links. Maybe even an Oracle gateway?
>

Interbase has/had an Oracle gateway. It hasn't been maintained
and isn't part of the main code tree, but I'm sure Ann would be
delighted to let anyone else have at it.

>3. Something like index-organized tables or cluster-index tables. I
>have a table with over 150 M rows (and growing) with non-unique index
>
>(typically about a 1000 rows per unique index value). I'm currently
>using Oracle's cluster-index table for this with very good results.
>I've copied the data to IB, but Oracle does have a performance
>advantage.
>

Do you have a case that shows this? Clustered indexes generally
exist to avoid excess disk rattling when traversing a table in
index order. Interbase doesn't use the index for ordering unless
the idiotic index traversal option is used and always traverses
tables in physical order, obviating the need for clustered indexes.
It would be nice to know what is actually going on.

>My point is not to start a debate on "this is how you should do it
>with IB," but rather how to use IB and Oracle side-by-side and start
>removing the management fear of non-Oracle database.
>

Some people will never feel comfortable with anything other than
Oracle and the U.S. Postal service. For organizations in which
it is better to fail with Oracle than succeed with anything else,
Interbase will be a tough sell. It remains to be seen whether it
will be a tough give-away.

Jim Starkey