🚀 go-pugleaf

Hi everybody,

I am currently evaluating IBReplicator to see if it can solve my
problem...

I wrote some years ago an application which uses Interbase as the main
database. Of course I didn't have replication in mind at the time.

The database is shared between 10-15 users over a LAN in one office.
Now they want to expand to a remote office in which there'll be
initially 2-3 users, but they could be more in the future.

Now, every table in the DB has its own numeric ID field mantained by
generators. In some tables the generator is incremented by a trigger
(i.e. before insert), but in other cases generator is incremented by
the application, because it needs to know the ID before the row is
actually written.

Since generators' values are not replicated, I have to modify the way
the application handles records IDs.

I thought I could build a IDs table and stop using generators. This
table would have only two cols: TableName(PK) and ID value. The ID
value for each table could be incremented inside a trigger or in the
application just exactly a generator and the table could be replicated
by IBReplicator. The problem arises when the connection between the 2
databases is down for a while. In this situation, any of the two DBs
could individually increment the value of the ID and when the
connection is back there can be a data loss.

Looks like the only solution is to use non-numeric random generated
unique IDs as primary keys!! ARGH!! This would mean a complete
redesign of database and a LONG series of modifications inside the
application! I'm considering the possibility to kill myself... :))

Has anybody ever happened to have this problem? I'd like to continue
using simple numeric IDs. Is there a way to achieve this? Otherwise,
is there any solution that doesn't terribly affect the app and DB
design.

Please help me. Thanks in advance,
Guido Pietrella.