Re: SQL, Logical Files, Unions & Indexes.

Hello, all:

1. with multiple libraries, you need only to manipulate the library list to ensure the 
library containing the desired "company" data tables is in the *LIBL.
2. with multiple members, you must issue OVRDBF commands for each PF each of 
the LFs used ...
3. with multiple members, you can specify MBR(*ALL) on the OVRDBF command to 
process ALL records in ALL members.

Could someone comment on the "pros and cons" of the above approaches, versus some of the 
other SQL "join" approaches mentioned in prior e-mails in this thread, from a performance 
standpoint?

Also, a more "pure" SQL approach would have just one large database table, with one member, that would 
contain all of the rows for all companies (or states, etc.), and add a field in the record for "company" or 
"state" (if such a field is not already there?).  Of course, this field would have an SQL index built over 
it.  Then, it becomes easy to select records for only one individual company (or "state" etc.), or select ALL 
companies, or any group of companies.

Would anyone care to comment on the relative merits and "pros and cons" of this approach, from a 
performance standpoint, versus the original approach of using multiple "sets" of PFs & LFs, and joining 
them, and versus the more "classic" OS/400 approach of using multiple members within the same PF & LFs 
outlined above?

Thanks.

Mark S .Waterbury