Correction:
....Since 2004 we're using Mediator with our Clipper applications and I
have to say that in my 30 years of IT development experience, I never
have
seen a tool that can keep up with the reliability and performance of
Mediator.
Congratulation to the folks at OTC. ...
Fred
On Feb 21, 6:55 am, d...@[EMAIL PROTECTED]
wrote:
> Hi fellows,
>
> I know from few years back (2204/2005) that the Jacek/Bozena from OTC
> in Poland were frequent reader of this newsgroup and I hope they still
> are.
>
> Since 2004 we're using Mediator with our Clipper applications and I
> have to say Mediator this a middle I have to say that in my 30 years
> of IT development experience I never have seen a tool that can keep up
> with the reliability and performance of Mediator. Congratulation to
> the folks at OTC.
>
> Since 2004, more and more applications (PL/SQL and Delphi) in our
> corporation using the tables on Oracle in concurrency to Clipper.
> At the same time, some table size have grown significantly.
> Now we're facing the problem that we have more and more errors on
> "Unique key constraints violation".
> For the non-Clipper applications we implemented "BEFORE_INSERT"
> triggers to populate the next available RECNO sequence via "Select
> max(recno) from <TABLE>".
> On tables with more than 12 million rows this is getting more and more
> time consuming and we have the suspicion that due to the execution
> time for this statement and the simultaneous use of RECNO inside of
> Mediator, we receiving the key constrain violation.
> We know that we could turn of the internal buffering of RECNO in
> Mediator, but this bring down the performance of all Clipper
> applications. And also we considered already the use "NEXTVAL" from
> Sequence for the triggers, But that doesn't solve the problem that
> Mediator has still the internal RECNO sequence.
>
> This is our dilemma; we need a high performance AND the reliability of
> a unique primary key.
>
> One idea during our brain storming session about this problem... If we
> could use a Oracle sequence for
> a) all Clipper apps and
> b) all non-clipper apps,
> we would have serialized the process to fetch the next value from the
> sequence (data dictionary related) AND the guarantee to have always a
> unique PK (RECNO) - What do you think about this idea?
>
> I hope some developers maybe had already the same problem and found a
> solution for the above described problem.
> And maybe the folks from OTC reading this posting as well.
>
> Looking froward to hear from you some feedback.
> Any replies are much appreciated.
>
> Fred
|
