We are currently running an application under Windows 2000 Advanced Server using
Terminal Services. Our clients connect to the servers using Internet Explorer on Windows NT 4 and Windows 2000 workstations. On the first connection, this will require the downloading of an Active-X component, so the user sufficient permissions to install the control or an administrator will have to handle this first connection.

The application is installed on the Terminal Server just like it is on a stand-alone computer, and any registrations required at setup will be handled at that time. Once installed, the server will run a separate process for each user. Because of this, your application should use the SYS(3050) function to limit foreground and background memory usage. Windows NT 4 workstations require a Client Access License to connect to the server (the license is included with Windows 2000 and Windows XP Professional) and the server monitors connections. If you have a limited number of conections, you can configure the server to disconnect idle users.

Some users exit their Terminal Server session by closing Internet Explorer. However, this leaves the instance of their application running on the terminal server, waiting for them to reconnect. This may not be desirable, and consumes resources.

Terminal server applications are easy to maintain, but not while they are in use. Generally, you will be able to perform normal maintenance chores while users are on the system.

As terminal services works by transmitting keystrokes and mouse clicks from the client to the server and returning bitmapped images, the performance can be very good, even over a 56K modem. If you have users connecting via modem, it is important to use a lightweight image as the background desktop (if you use this type of image), as it is the image rendering performance that limits performance over narrow bandwidth connections.

It is important to pay attention to housekeeping chores (closing unneeded cursors and tables, using arrays instead of cursors or tables when possible) as one can unexpectedly run out of file handles. The default number can be increased, but you don't want to find the limits the hard way. :-)

The terminal server stores the path to the C:\Documents and Settings\UserName folder in the USERPROFILE environment variable, and this can be retrieved using the GETENV("USERPROFILE") function. Use this path to store any user specific settings. It may require some re-writing of your application.

Likewise you can any of the other environment variables in the same way. USERMACHINE stores the computer name of the workstation that the user is connecting from and TEMP stores the path to the user's TEMP folder in the Documents and Settings folder. What's nice about this user temp folder is that the terminal server will automatically delete anythin in it when the user logs off the server.

Hope this helps,

John