I appreciate that with PL/SQL being rather more effective than, say writing
cursor based code in SQLServer, which is my background, I may be failibg to
compare like with like. I'd like to ask a general question on best
practice, though..
Now any any other RMDBS, at least I've worked with (SQLServer, SQLAnywhere,
Informix, Oracle RDB) using a cursor is always a horrible performance hit
compared to using a well structured set of views, if required, accessed by a
stored procedure (Procedure).
As I understand it - while PL/SQL kills SQLServer in terms of optimising a
transactional approach - compared to a more relational setup - PL/SQL s
still translated into set based queries - so using it, if it's easily
written using views and Procedures, using a lttle common sense and taking
advantage of parallism - I'd guess, in general - usind cursors inPL/SQL will
will take moe server resources and - basically - be less efficient.
Seems where I'm working that using PL/SQL is used to mimic IDMS - and while
I know little about Oracle - I'd be surprised if this approach is optimal -
certainly is a killer on any other RDBMS I've worked on.
Am I missing something? does anyone have good links on the benefits (or
otherwise) of the two approaches?
Sorry it this isn't clear - the reasoning behind some of the use of cursors
to mimic joins is way beyond me. but I'd always been given to believe that
use of cursors was a total last resort, not a good standard approach.
Anyway - cheers for your input. May your children all be jockeys who
permenantly ride winners.
Received on Tue Jun 06 2006 - 17:07:18 CDT
