I'm trying to replicate (transactional, push) a relatively small table (50K rows), subject to several different filters, which are rather expensive (each taking at least an hour long to complete, involving a join with a table having 4 billion rows in another database). I expect that the published article will have to be synchronized a few times a day.
Is this feasible, or should I look for a different solution?
- Will there be any issue with exclusive locking of the original table during replication?
- Currently, the Snapshot Agent status says "[0%] The process is running and is waiting for a response from the server." Is this normal for an expensive filter? Is this acceptable in practice?