I am using Sync Framework version 2.0 in a collaboratin scenario sync'ing between two sql server databases. I am initially provisioning my "client" database using an initial sync, and it is taking a long time (> 30 minutes) and it is not a large database.
I started playing with various options, and I discovered some strange performance characteristics: when I switched on batching I was able to watch the batched files being created on the server and the client. On the server the batched files are created quickly, but on the client they are created approximately every 20 seconds, and each file is about 70K in size. When monitoring the connection between the client and server machines I could see that the T1 that connects them is saturated with the traffic between them and the only communication between them is the sync framework.
Since a T1 should be about to manage a 70 k file in less than 0.5 seconds ( as opposed to the 20 seconds I am seeing) I am concerned that something is wrong either in my setup or in the framework.
When I run the same scenario but with the sync framework server running locally on my machine (so I am making a Sql connection from it to the remote server) the test executes much, much faster, which indicates to me that the communication overhead is coming from the framework.
Has anyone seen the same problem, and/or does anyone have any suggestions?
Nick Caramello
