Re: [Hampshire] Steve Gibson (Was: Re: Data Recovery)

Top Page

Reply to this message
Author: Vic
Date:  
To: hampshire
Subject: Re: [Hampshire] Steve Gibson (Was: Re: Data Recovery)

> James, I was going to ignore this post but I was concerned that this
> misinformation would be to the detriment of Stephen


The trouble with dealing with "misinformation" is that you need to be
absoutely positive that that's what it is...

FWIW, I think James was a little harsh, but he's not completely wrong...

>>From http://www.grc.com/srfeatures.htm
>
> "Direct Hardware Interaction with Hard Disk Drives and Controllers:
> Unlike any other disk utility, SpinRite interfaces directly to the hard
> disk system's hardware, rather than working through the system's
> operating system or BIOS.


You see, this bit is disingenuous; although it might be addressing the
drive unit directly (rather than relying on the BIOS), it's very unlikely
these days to be addressing the heads & motors directly.

Modern HDDs are actually embedded computers. There is a processor on board
to manage the flow of data from interface to platter. When I worked for
ST, pretty much the whole of the ST10 production went to disk drives.

This has some surprising effects - for one thing, some drives deal with
prefail conditions by having some "spare" sectors. These get mapped in by
the firmware when necessary. This increases the MTBF, but means that a
logical-to-physical mapping must be kept. It also means that trying to
deal with the disk like we used to is not going to provide the results you
might expect - because the controller card no longer gets a say in the
physical placement of data, just in its logical representation.

> This allows SpinRite to immediately determine
> the location of any surface defect after a single incident of error,


Can you see why this no longer holds?

> SpinRite selectively disables and enables a drive's
> read caching, write caching, read-ahead buffering, on-the-fly sector
> relocation, on-the-fly error correction, dynamic servo thermal
> re-equalization, early and late ECC error correction, and other
> advanced features in modern drives.


All that sounds pretty nifty - and TBH, I'm too long out of that game to
know whether it bears any resemblance to the truth or not.

> Not at all. We sell many copies of SpinRite every single day to the
> many people who are having serious trouble with their modern drives.


This bit is just fluff - the fact that you can sell snake oil to someone
who is desperate enough to try anything means nothing at all.

Vic.