There is nothing in Windows that requires an application to be installed. That said, installation is intended to:
- Make things more simple for the end user.
- Setup the registry, usually for path information and uninstall information.
- Initialize any initial information the software may need before it's first run.
Simply avoiding using the registry and saving files locally to your application is usually enough to make your application portable.
That said, as long as you allow the user to select a database location within your software, you should be fine. Saving the information on the pen-drive, in an .ini file for instance, would allow each computer you plug into to read these same settings.
If you expect each computer to have a difference connection string to the database, you could save your settings to the %appdata%
directory. When the user plugs the pendrive back in later, his settings will still be there, and no other user will see these same settings.
The downside to the second approach, however, is that the user has no way to "uninstall" and recover the space written to %appdata%
automatically. However, for most private business applications, this isn't much of a concern.
Edit: If your real question here is how to distribute an application without an installer, simply build the Release version of your application, and look in /bin/Release/
within your project. Copy these files to another location, remove any debug or unneeded files, and make sure you have all your dependencies in order.