We run about 750,000 transactions per day. For adequate performance, we
find that it is necessary to move rows that are more than 2 weeks old
into archival tables in the same database, and to segregate real time
transactions from ad-hoc analysis into separate databases.

On Thu, 2004-10-14 at 05:06, Planles wrote:
>
> > So my answer is : why, when running a RDBMS, do you need to
> > archive at all?
> > We run at about
> > 30 000 transactions per month in the log table, and we have no problems
> > with that kind of
> > transaction density.
>
> Basically just because of hidding old orders from user.
> So you suggest, that I should keep ALL orders in one table and omit
> accessable one with select statement e.g.
> select * from orders where OrderDate < some_date ?
>
> Regards,
> Primoz
>
>
>
> >
> > >Hi,
> > >
> > >Could you share with we, what expiriences do you have with
> > making archives
> > >of some table data ?
> > >
> > >Let me be more specific:
> > >in an application with clients, orders, employee an similar stuf
> > there can
> > >be a few houndreeds or more orders each year.
> > >Normaly we do not need to have all orders in actual orders table, but we
> > >should be able to check any old order at any time.
> > >So old orders should be transfered to another table with the same data
> > >structure as actual one.
> > >
> > >What approach do you prefer ?
> > >Make just a new table within same database (fdb file) or is
> > better and more
> > >secure to create separate database (fdb file) just for archive.
> > >
> > >I don't have in mind backing up my data, but exclude old orders
> > from actual
> > >orders table.
> > >
> > >
> > >Regards,
> > >Primoz
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >Yahoo! Groups Links
> > >
> > >
> > >
> > >
> >
> >
> >
> >
> >
> > Yahoo! Groups Links
> >
> >
> >
> >
> >
> >
> >
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>