Thanks for that. Could give an approach to a different problem we get occasionally where at some peak times we get bigger latency on our logreader agent - looks like...
The option to do the non locking snapshots as has been suggested helps as it means that while the main bit of the copy is working the table is available...