What happens behind the scenes on a data pipeline?

[Irene Fink]

I have a need to replicate changes from one database to another. I
had been using a pipeline with a delete/insert option and thought that
it simply deleted any rows found in the destination database that were
different, and reinserted a new record from the source database. With
testing, it looks like ALL the records are deleted from the
destination database and then reinserted. This presents a realtime
problem if I want to allow database access during my update process.
If a user queries a table at the wrong time they receive "no rows
found"... I tried testing the data pipeline using the update option,
but I realize the behavior of update doesn't seem to work well for me
as it takes twice as long as the delete/insert pipeline. Too, only
changes from the source table are propagated to the destination. If I
should have extra rows in my destination table, they are left in
place, so in the end, I do not end up with the same records in both
source and destination tables. Is there any other option or method to
do my table refresh that would allow users reliable access while the
job is running and insure that the two tables match at the end?
Thanks in advance. Irene