Subject Replication and redesigning the wheel
Author Gerhardus Geldenhuis
I have a very simple replication mechanism in place.
Basically all my tables have a unique number field in the
table and all insert/edits gets logged together with the
table name. I don't allow deletes on any tables so that loop-hole
in my design is covered. I have not yet written the program that
interprets the logfile and wanted to ask for advice on how to handle
errors like db's that dont connect and what you must do when the
connection fails in the middle of an update etc.

In principle I am against redesigning the wheel(If the wheel are round)
Do you think it would be worth my while to rather learn the replication
components and use the nt service componentsof ibo than doing it myself.
Your immediate answer may be yes, but my reasonig wheter sound or not was
I have a million things to learn and little time to do it in. And because
my replication needs is fairly non-complex I thought that the complexity
of the IBO components is/was overkill for my needs.

Additional Info
Most of the replication is one way with only one or two tables that is
two way. We have a main "server" with two branches at the moment but
must allow easy expandibility to more brances.
All databases will be running on Linux and the linux box at the main
branch will monitor for tcp/ip calls to the branch databases and then
do dialup networking(with pppd)