>Does that include using Sql Server (and if I can use remote views in VFP)? Is Sql Server the server technology required, and terminal services the transport technology?

These two independent questions.

You can use SQL server from a remote machine, and just recheck the connection before sending your updates, or have some code that will handle the case where the connection has gone south, by reconnecting and repeating or giving up after a given number of tries. For connecting to a remote machine you don't really need anything else but its IP address, and username and password (the ones defined for the database's users, not for the machine login), and of course the rest of the connect string. E.g.

	lcConnectString ="Driver={SQL Server};Server=xx.xxx.xxx.xxx;UID={username here};PWD={password here};Database={db name here}"
this.nGeorgFridrihHendl=Sqlstringconnect(lConnectString)

For the rest of operations, you always use the handle returned by SqlStringConnect(), for each SqlExec() call. That's if you're using SPT cursors. For remote views, you can create a connection with a connect string like the one here.

>Can I use terminal services with plain VFP database tables, or is the front-end processing too much for this setupu?

Nowadays TS is just cool. Haven't used straight TS though, my experience is with Citrix (which is something like TS on steroids, to my knowledge - someone may chime in with more detail), and it was just as simple as running the app locally, except that the only thing that travels over the wire are keypresses, mouse actions and screen updates. The data remain on the server(s).

One trouble I had at times, while running a VFP exe under Citrix, was that if connection breaks, the app isn't aware of it. Citrix does kill it after a certain time of inactivity, which was set too large IMO (about 15 minutes). Sometimes I'd be able to run the same app and actually reconnect to the instance already running; most of the time it'd just flatly refuse, saying the app is already running. Nothing bad happened, though, the tables remained intact, so I didn't have to try to have the app shut down before this period expires. Probably the systematic use of transactions and FLUSH after each tableupdate() made sure nothing happens to the tables.