Re: SQL, Logical Files, Unions & Indexes.

[Evan Harris <spanner@xxxxxxxxxx>]

HI there

Well I guess if you have to segregate your data into separate 
libraries (and a reasonable explanation for this is separate 
companies wanting pseudo physical separation of their data) then why 
not run separate queries for each library - possibly even in parallel 
- and then aggregate (or whatever) the data afterwards ?

My experience has been that you don't generally want random access to 
the files in different libraries, but when the need arises it is 
usually for extraction and creation of statistical and summary data.

I grant you that this probably needs some additional work up front to 
set up but might well be a good/better performer.

Maybe if you could give me an example of a task that required you to 
read the whole table as a single query I could see what you're getting at.

regards
Evan Harris

At 08:53 a.m. 2/04/2007, you wrote:

> Birgitta,
>
> in the original message Mike talked about an Index over a series of
> analogous tables in different libraries.
> He stated that it's not possible to create an index over all the tables, and
> I don't know of any way either.
> (or is there ?) .
> As you mention "with the right indexing strategie SQL performs much better
> than natibe I/O" .
> Therefore it should be possible to create indexes over multiple tables, just
> as it is possible to make logical files over multiple files? .
> (I cannot see how views, without indexes, are more powerfull than  logical
> files)
>
> So the only solution at the moment -as I see it, the same that Mike
> suggested,-
> is to create a DDS described Logical file over all these tables (or PFs).
> In this case I don't see another way but
> "A DDS described logical file has to be used in the SQL-statement ".
> and therefore
> "with the logical (the one over all the tables) the new SQE will not be
> used" .
>
> Any suggestions or alternatives are more than welcome.
>
> luc