RE: Line Item Budgeting

To second Walden's assertion and add some SQL challenges with multi-member
files (has been discussed before):

- you can't build SQL indexes over secondary members. So, if table scans are
not desirable, you'd have to make sure to build keyed LFs as a substitute.
No big deal you say, but yet another thing to add to the to-do list
- you have to create ALIASes or use OVRDBF to access secondary members in a
table
- you can't create SQL views over secondary members (just no way to swing
this one, so it's a big minus, in my book anyway).

There are other ways to partition annual data. For example, you can use
CREATE TABLE AS support to create a file with a particular budget year for
archival purposes (then clear the base table of that year's rows).
Multimember files work fine for native I/O and OS commands like CPYF and
like, and as you say, even offer some functional and performance advantages.

But for SQL.... they're a headache to say the least.

Elvis

Celebrating 11-Years of SQL Performance Excellence on IBM i, i5/OS and
OS/400
www.centerfieldtechnology.com


-----Original Message-----
Subject: RE: Line Item Budgeting

Forget multi-member, it's not worth it.

SQL performance would be enhanced by dividing the table into members.

Why? SQL has no problem with many rows, why would you believe that SQL
would be faster. OK, I guess in _theory_ there are fewer rows, therefore
fewer reads, therefore faster, but in reality you can't tell the
difference.

Back-up and restore procedures would run quicker by simply backing up

the *last member.

Really. Seems to me simplicity is the king of backup/recovery. I can see
them constantly backing up *last, and then needing to restore the entire
table after a crash and saying "tapes from last year, nope... can't we
just use last nights."

Actually I'd think your programming would be simpler too as you could
simply have a key for the year and then SETLL or SELECT as needed.

The copy last year to this year is a trivial program, using that as a
reason to go multi-member is not a valid argument.

-Walden