Usually what kills option 2) is all the network traffic generated by
retrieving intermediate results. Option 1) is done server-side and only the
final result is retrieved. Also, if the same table is joined more than once,
option 1) can use buffer cache more efficiently.
<ctcgag_at_hotmail.com> wrote in message
news:20021211011610.461$9S_at_newsreader.com...
> "O.C." <clubmitglied_at_yahoo.com> wrote:
> > Hi !
> > What is faster in embedded SQL (in PRO-C):
> > 1. One join of 5 large tables or
> > 2. Splitt the this big SQL into 3-4 parts (and use the output of one sql
> > result
> > as bind variables for the next sql statement)
> >
> > Indexes are in use.
>
> If the CBO is working well, number 1 should almost always be as fast
> or faster. Otherwise, number 2 may be faster (by orders of magnitude).
> Use #1. If that causes major problems, then (get a DBA to) tune the
> query. If that fails (or you can't wait for an overworked DBA to get
> around to helping you), try #2.
>
> Xho
>
> --
> -------------------- http://NewsReader.Com/ --------------------
> Usenet Newsgroup Service New Rate! $9.95/Month 50GB
Received on Wed Dec 11 2002 - 10:53:52 CST
