Re: ProFile HD setup

[mjmahon@aol.com (Michael J. Mahon)]

Ernest (>) replied to me (>>):

>> >The whole process only takes 30 seconds or so.
>>
>> Something is fishy here.  10MB is 10,000KB, or 20,000 512-byte blocks.
>> How does a 1980's disk drive accesses 20,000 of _anything_ in thirty
>> seconds?
>
>Psst! You ever see "Willy Wonka, and the Chocolate Factory?"
>
>Well, change that to "Steve Jobs, and the Apple Factory" and you'll begin to
>get the idea. Apple was able to do all sorts of strange and mysterious
>things, that science cannot begin to understand... or replicate.

LOL!

>> And what does it mean to "relocate" a bad block?  If the block is in use
>> and cannot be correctly read, then although the block can be re-mapped
>> to a spare, its content will be wrong.  This is not exactly graceful error
>> recovery!  Controller-driven sector sparing makes sense 1) on a blank
>drive,
>> prior to OS formatting (since data content is irrelevant), or 2) on an
>in-use
>> drive if retries (with possible threshold tweaks and/or error correction)
>can
>> recover the data intact.  In the former case, it does not make sense to do
>> it each power cycle, and in the latter case, it does not make sense to
>> _wait_ for the next power cycle.
>
>See, it's like this. If you move the damaged block to slot 2, and replace it
>with a fresh one from sector 1, then add, sector parsing  to the bit mapper,
>the repairs and block scans are faster, and more fault tolerant. It's simple
>really.

Actually, Earnest, I understand about substituting spare sectors--even
how to structure it so that latency is not increased!--as you can tell
from my post.  The issue is when a controller, not knowing anything
about the "significance" of the data in any sector, can relocate it to
mask an error.  The issue of lost data stems not from the controller
not retrieving the physical sector mapped to the logical sector address,
but to the controller being unable to retrieve what may be significant
data from the damaged sector in the process of relocating it.


>> I'm not saying it doesn't do it--stranger things have happened--but I am
>> saying that it doesn't make sense.


-michael

 Email:  mjmahon@aol.com
 Home page:  http://members.aol.com/MJMahon/