How would this affect replication? Would replication flow smoother than doing a delete * from the large row table?
If you are planning to delete large numbers of records and replication is involved, I would implement the logic in a stored procedure and replicate the action of that stored procedure. As part of the definition of the article for that stored procedure, tell replication to replicate the actions for the stored procedure.
By default, replication only replicate the definition of the stored procedure. When you execute the stored procedure, the results (the deletes) are recorded in the transaction log as individual deletes of each record in the table being deleted. The deletes are "copied" from the transaction log into the distribution database and then applied to the subscriber.
By replicating the action of the stored procedure, the stored procedureitself is executed on the subscriber so, instead of replication having to execute several million DELETE statements (including storing them in the Distribution database), replication will simply execute the stored procedure in the subscription database.
This technique reduces the size of the relevant records in the Distribution database and reduces network traffic between the distributor and the subscriber