I'm confused by site boundaries versus roaming boundaries in SMS 2003. What is the difference?

The simple answer is that site boundaries apply to the legacy clients, and roaming boundaries apply to the advanced clients.

Let me explain.

Legacy clients (Win 9x or NT4 with the CAPINST or SMSMan client installed) do not have the ability to roam. In fact, if they find themselves on a subnet other than the subnet in which they are assigned then they uninstall the SMS agent. They are completely controlled by site boundaries. In other words, they act exactly the same with SMS 2003 as they did with SMS 2.0.

Advanced clients (Windows 2000 or later with either CAPINST /ADVCLI or CCMSETUP installed) do have the ability to roam. When they roam, they can roam to either a well-connected subnet (like a LAN) or a not-well connected subnet (like dial up, VPN or wireless). The local roaming boundary comprises all the subnets that are well connected, and the remote roaming boundary comprises are the subnets that are not-well connected. You need to differentiate between these two types of roaming boundaries so that you can control how software will be deployed when the advanced client is well connected to the nearest distribution point versus not-well connected to the nearest distribution point.

So when you are configuring SMS boundaries, enter the site boundaries as the well-connected subnets (this will sort out your legacy clients, assuming you have no legacy clients using dial up, VPN or wireless). Then accept the setting to set the local roaming boundary to be the same as your site boundary (this will sort out well-connected advanced clients). Then assign your dial-up, VPN and wireless subnets as being in the remote roaming boundary (this will sort out not-well connected advanced clients).

Now everything is correctly bounded. Bliss.

Regards

Paul Eddington

KB Keywords: SMS, Systems, Management, Server, site, boundary, local, remote, roaming, boundary.