We had a bit of an issue with some of our remote clients. Several locations with low bandwidth (less than 1Mbps) were competing for space on a T1 connection back to the data center. After the initial install all the clients were trying to get their policy and pull definitions down. This maxed out the connection back to the data center and caused the install to run hours past the window.
Look at your network architecture and see what the bottlenecks are? Do you have multiple sites that share a single connection?
Consider a Group Updater for remote locations.
Create a policy that uses the Group Updater. Use this policy in your install.
Perhaps a build with the latest definitions incorporated.
Once everything was installed an up to date, network bandwidth usage has been minimal with no noticeable spikes. It was just the initial roll-out. I think we took on a bit much at once not realizing the network architecture and bottlenecks that were in place.
Hindsight....