Subject Re: Speed of Firebird v1.52 vs. Interbase v7.1
Author Adam
--- In firebird-support@yahoogroups.com, "Steve Harp" <steve@h...>
wrote:
> --- In firebird-support@yahoogroups.com, "Alan McDonald" <alan@m...>
> wrote:
> > > Hi All,
> > >
> > > I have switched a client's application from Interbase v7.1 to
Firebird
> > > v1.52. I've made no other changes to the application except the
> > > database server. The application uses IBO for connectivity.
> > >
> > > One of the users claims that the application is now slower.
Could
> > > there be any truth in this? I've never heard of Firebird being
slower
> > > than Interbase.
> > >
> > > Thanks,
> > > Steve
> >
> > easily done if your indexes are wrong/don't match the optimizer
> plan. If you
> > went the other way it's possible Interbase is slower than FB for
the
> same
> > reason
> > Alan
>
> I didn't change any indexes. The way I create the FB database was
to
> generate a script from IBExpert using the Interbase database and use
> this to create the FB database. Then I backed up the Interbase
> database and used Firebird to restore it into a temporary database.
> Lastly, I copied the data from the temporary database into my new FB
> database.
>
> The FB database should be an exact duplicate of the Interbase
database.
>

Hi Steve,

I think what Alan was trying to say is that Firebird was based on a
beta of Interbase 6. Both Firebird and Interbase have subsequently
improved their optimisers. It is possible that there are some queries
that the Interbase 7.1 optimiser handles better than the Firebird 1.5
optimiser, and vice versa. I suppose Borland has the upper hand here
as they are able to see the enhancements Firebird has added and do
the same sort of thing if they deem it worthwhile.

It is no secret that there are better optimisers than in Firebird 1.5
(Firebird 2.0 for example), but what Alan was saying is that had you
built your application on Firebird 1.5, then ported it to Interbase,
you may well find Interbase slower.

On the bright side, you will probably find it is just a few queries
that are no longer using a sensible Plan. Identify them and force
them to use the better plan and there should be little if any
difference.

Btw, I am assuming you are running them on the same hardware so we
are talking apples and apples.

Adam