You have a webserver that needs to access files on shared storage. At some point down the road, the files may move to another, bigger server with a different name. Which of the following would be the best way to configure your Microsoft Windows NT/2000-based Internet application to read those files?
A) map a drive to the storage and include the path in the scripts
because you are unaccustomed to working with a server. When you logout of Windows NT or 2000, drive mappings disappear: they are associated with a particular user account. Servers sit in rooms by themselves, and, if the people who own the server are not entirely clueless, will not have an active login session open on the console. Answer A is fine if you're a schmuck. (I believe the same holds for answer C but I haven't confirmed this in practice.)
You could answer D, and say that you'll use a dual-homed SCSI controller. This is somewhat expensive, though it worked very nicely in practice with DIGITAL Clusters for Windows NT. You could also use a FCAL SAN device.
But why bother? Microsoft has provided you with the ability to access remote files from within Internet Information Server: just use the Universal Naming Convention (UNC) path to the files on some other host, and as long as the permissions are hunky-dory, you're in like Flynn. The correct answer is B. Now you too can be an MCSE.
(Originally published at schmuck.editthispage.com on August 13, 2001)