Install to external drive: good or bad idea

Question

I have an iMac that is currently on 10.5.5. However, it was originally a Tiger install. I performed an update install to 10.5.6 and while it runs, I have stability issues that came with the Tiger -> Leopard update that I know I can eliminate by doing a clean install.
Of course, the problem with a clean install is that the total execution time is very lengthy (from install to last app license re-registered) and my workstation is DOA (or nearly so) until completion. And then, there's always the odd detail I forgot about that I have to research to find the answer.
So, why can't or shouldn't I do the following, or how to make it better?
1) Do a clean install to an external drive (I have a 500 GB USB drive I can use) Leave the internal drive intact until step #4.
2) Install the OS, apps and utilities until it matches my current machine.
3) Once ready, back up the internal drive to the external drive as a DMG. (BTW: there will be another time machine BU on another machine on the network -- so there will be redundancy.)
4) Do a secure 7-pass erase of the internal hard disk (this performs hardware-managed low-level sector checks, turning off sectors that error on read/write)
5) Use Carbon Copy Cloner to copy the external data to the internal boot disk
6) Copy my documents from the backup to the new user folder.
7) Viola! Reinstallation complete but I can access my data at nearly any time (latency depends on current task) if I must (except during #4)
Where's the fly in this ointment?
TIA!

Answer 1

However, is there a way of achieving zero-ing out defective sectors, like SpinRite, <http://www.grc.com/sr/spinrite.htm> without my having to remove the hard disk?
AFAIK, SpinRite does not actually "zero out" defective sectors in modern IDE hard drives. To better understand what it does (& in general the link between analog & digital data reading mechanisms in magnetic media), look beyond the brochures & study the “What's Under The Hood” technical note at http://www.grc.com/files/technote.pdf, especially the "Surface Analysis" section beginning on page 9.
SpinRite's makers want you to believe that "weak" sectors are something to be very concerned about, in effect that reputable drive makers design their modern high density drives (ones several generations beyond those using MFM) to regularly rely on these 'defective' sectors to boost capacity, & that only SpinRite's technology will map them out before they cause data loss. This is a dubious idea at best. The techniques used in SpinRite's surface analysis are essentially the same ones modern ATA drives use to monitor for "grown defects" -- except that modern ATA drives contain their own highly capable, purpose-designed proprietary hardware & dedicated operating systems that SpinRite can access only indirectly through clever use of standardized extensions to the ATA command set to do what the drive does anyway.
This is not to say that the product is without merit. It is primarily purposed as a data recovery tool, & for this the extended, low level statistical analysis of the 'raw' analog data from unreliable sectors is a powerful (if not unique to this product) method of data recovery.
There is also some merit to the idea of proactive early detection of unreliable sectors that if used for data storage would eventually require the drive to struggle & perhaps to fail to get 'clean' data from them. (I believe this is what you mean by "zero-ing out defective sectors," just in more technically precise terms.) However, be aware that drives already do this, just with actual data instead of specially designed signals, & in general this is enough to detect sectors that are going bad before the data becomes unrecoverable, at which point the drive moves the data to another sector & maps out the first one to prevent its future use.
In real world terms, these unrecoverable 'hard" errors due to grown defects are very rare in healthy drives, only becoming problematic in drives that are dying, typically because of head crashes, exceeding operating or storage temperature limits, or other traumatic events that affect the magnetic media layer. So something like SpinRite's surface scan is most useful for extending the life of a drive that has already become unreliable or for prescreening drives for gross manufacturing defects. Since the first use makes little economic sense in these days of cheap hard drives & the second use is potentially superfluous because gross defects would likely be quickly detected anyway, users must judge for themselves if using the product for this is worth it.
(The same considerations apply for deciding if or when the frequently seen user advice to 'zero out the drive' as part of preparing it for use accomplishes anything of real worth or not.)
Finally, getting back to the idea of moving the drive to a PC to run SpinRite & why I think it doesn't actually "zero out" defective sectors, I can think of no way it can force the drive to actually map out a sector unless the drive itself determines it should do so. This leaves the alternative of marking the sector in use in the file system's directory structures, which in effect would accomplish the same purpose, but with one important exception: its effectiveness is dependent on the file system remaining in place on the drive. IOW, if a user changed the file system of the volume or changed the partition scheme, the file system metadata marking the sector as in use would be lost, & SpinRite's surface scan would have to be done again after these changes to get it back.
I want to make it clear that I am not certain that this alternative, file system based method of achieving what Spinrite claims it can do is actually how it works, but I can find nothing in the product's documentation that contradicts or conflicts with it, nor am I aware of any standard extension to the ATA command set that would allow SpinRite to directly set the grown defect list. If someone has info that shows I'm wrong, please post it. I have no problem abandoning theories that don't pan out.
Message was edited by: R C-R