It is a matter of perspective. Someone has to walk before they can run. Tedi1 needs something that will work. In a perfect world every one takes an image backup of every box and in case of a failure has another identical piece of hardware to reimage the whole server onto over a wideband network. The LAN guy is ready to drop everything and the whole restore is done in 1 hour on a spanking new machine that was sitting in the closet.
In the imperfect world I see every day, another machine identical to or with the same capacity as the first is never available, no one reads installation instructions, the LAN guy cannot get around to locating the backup tape for two days, and when he does it takes 18 hours to restore 4 gb because of a slow backup device and the network is busy. And half the time some hardware failure occurs or something has to be jiggled to get it to work right.
Everything you say may be true, but it is equally true that Tedi1 is Oracle challenged and he needs something that will work RIGHT NOW. Granted you need a working db to use imp - but the whole purpose of exp/imp is to backup the DATA - if he reads the book then he will know he has to also backup some other files as well.
First things first. Do the export and the data is backed up right now. He does not have to go thru a long learning curve to do that. He can move on to more complex backup and recovery methods after he reads the book and someone trains him. But in the meantime he has a tool that can operate at the macro level of backing up all data incl database system schema level, but also has the granularity to restore just a table or a portion thereof without the need to get the LAN guy and the UNIX admin involved in a restore of the files shipped off to Iron Mountain.
In the imperfect world I live in - which is a development software company - we just have development databases and no 24 x 7 production systems. In this world - export and import works just fine. Yes - programmers lose tables- and I can get them back up and running with my db exports in 1-2 hours - whereas the LAN guy with his image backups has to order the tape, search the tape for a couple of hours and maybe in 24 to 72 hours a datafile can be restored - but he does not have the granularity capability to restore the missing table that someone on a tight schedule for devlopment is desperate to get back up and running. With archiving turned on I can restore right up to the moment of the loss - the image backup can only restore to the last date of backup.
Given that Tedi1 describes this as the only Oracle db is a sea of SQLServers, I would venture this is a somewhat smallish db with low traffic that is not necessarilty mission critical - and likely does not have the 32k rowsize you talk about - otherwise a DBA would be under contract to keep tabs on it. With these assumptions, the exp/imp and backup of the control files and init.ora would probably be sufficient on a regular basis. If something drastic occurs where the machine is lost - the local disaster recovery planning should kick in for that kind of recovery. If needed, they can get a contract DBA to come in for a day or two and create a new Oracle instance - and rebuild the database - with or without the control files and init.ora -from the imp file. Given the circumstances - I would suspect the management would probably will go for the low-budget solution.