[rescue] Total corporate madness (
Charles Shannon Hendrix
shannon at widomaker.com
Wed Aug 6 10:23:38 CDT 2003
On Mon, Aug 04, 2003 at 12:57:22PM -0400, Dave McGuire wrote:
> As I understand it, serial ATA implements most of the SCSI command
> set over a high-speed serial link.
I think they've gradually been trying to SCSI-ize IDE for years.
> ...which makes me wonder...why not just use FC-AL? It implements the
> SCSI command set over a high-speed serial link. Sound familiar? ;)
> And FC-AL is standardized, chipsets area already out there, it's
> scalable to VERY high speeds, etc.
Yeah, wonder is the word. You have to wonder why they keep pushing a
brain-dead interface when mass production would have made a much better
interface cheap, years ago.
> I think their point was to have "ATA" in the name and be able to say
> "it's not SCSI" because the PeeCee weenies are being trained that SCSI
> is somehow bad because it's more expensive than ATA
No, not really. It's still very ATA. For one thing, I believe you can
only have one drive per wire.
Might be a good thing really, since from what I have noticed, an IDE
drive dying usually takes the bus with it. So, if you want reliable IDE
raid, you need to stick to a single drive per controller.
The exception I suppose is special raid controllers which (hopefully)
don't allow any one drive to lock up the bus.
> Read about FC-AL, aka FibreChannel. That's essentially what it is.
> It works, it's SCREAMING fast, it's standardized, it's supported, and
> it's readily available.
But... its kind of a catch-22. It's expensive because not many are
sold, and not many are sold because it is expensive.
There is immense momentum behind IDE, much of it marketing and
UNIX/Perl/C/Pizza____________________s h a n n o n at wido !SPAM maker.com
More information about the rescue