Absolutely agreed. CPU is there to be used after all. :) Public synonyms
make developer's life so much easier I am willing to give away a few
more CPU ticks and memory to scan them and bring them into library
cache. Tanstaafl...
--
Vladimir Zakharychev (bob@dpsp-yes.com) http://www.dpsp-yes.com
Dynamic PSP(tm) - the first true RAD toolkit for Oracle-based internet applications.
All opinions are mine and do not necessarily go in line with those of my employer.
"Nuno Souto" <nsouto_at_optushome.com.au> wrote in message
news:dd5cc559.0206161606.11ad4545_at_posting.google.com...
> yong321_at_yahoo.com (Yong Huang) wrote in message
news:<b3cb12d6.0206160926.25dc8eea_at_posting.google.com>...
>
> > Steve shows that synonyms, particularly public synonyms, are indeed
> > bad in terms of CPU usage. But he said at the beginning "the best way
> > to refute such scepticism [i.e. synonyms are bad] is with a
> > reproducible test", as if his following test would show otherwise.
>
> Note the difference between shared pool latch for "private synonyms" and
> "no synonyms" is much smaller for soft parses (the ones most likely to happen
> if your app is reasonably well designed). Public synonyms may indeed have
> an impact, but private ones may not. Steve also mentions a few caveats
> about the measurements and the load characteristics.
>
> Personally, I've tried both in 8.0 and found no appreciable difference
> in the type of applications that I run, with the load that I have.
> So I use them: they are darn useful for security and ease of access
> to a schema by multiple users. Will gladly pay the extra CPU
> I may incur for the practicality of using them.
>
> As in anything, testing is the most important criteria. Just because
> one site mentions a warning about public synonyms doesn't mean that
> EVERY database out there cannot use them. Each situation must be tested,
> rather than applying "blanket decisions".
>
> Synonyms in Oracle have a purpose and a well defined advantage for managing
> object access in a database, which we may have to pay for in terms
> of additional CPU use. This is one case where I don't mind "wasting"
> CPU if I can afford it. It's desirable functionality, that makes
> management of access much easier. IMO, it's a desirable "waste".
>
> Of course, if we are dealing with a 10**6 TPS benchmark system, we
> definitely do not want to use ANY synonyms. And we go for the max
> performance possible. I have the feeling that is not the case with
> the majority of databases out there.
>
> Just my $0.02 anyways.
> Cheers
> Nuno Souto
Received on Mon Jun 17 2002 - 05:31:51 CDT
