I have a well-connected remote site that I'm delivering snapshots to twice-daily of certain tables. Snapshot replication is the only method that I can use based on the schema limitations of the underlying application.
The snapshot is published via UNC path and FTP, and is about 6MB so fairly small. I'm not 100% whether FTP or UNC is being used actually. It takes about 5mins from the time the snapshot is compressed into the snapshot.cab file to when the replication monitor shows me the first of the scripts being applied, so I assume that gap is transfer time. I'm thinking its UNC/SMB given that it takes a lot less time to copy that much data over our 3MB connection via FTP.
It takes about 25 more minutes to load the snapshot, which seems like a lot of time for a modest amount of data. Lots of rows (several tables have 100K rows), but the actual amount of data is pretty small.
The subscriber is running SQLExpress 2005 on a 2.8GHz ProLiant with 3GB RAM that's otherwise not busy. I haven't done extensive profiling yet to see if there's disk contention or what else might be going on.
I'm wondering what the likely rate-limiter is here and if there's a way to speed this up at all. 30mins seems like a long time to copy and load 6MB of compressed data.