Messages in this thread | | | Date | Sun, 13 Feb 2000 14:54:42 -0500 | From | Sandy Harris <> | Subject | Re: Scheduled Transfer Protocol on Linux |
| |
Bjorn Wesen wrote:
> I don't doubt the feasability of the HW design at all at that price. The > main problem is probably simply cold feet. Imagine you are a hot-shot HD > manufacturer. You instinctively feel "this is cool stuff", and prepare the > design and R&D. All is well. Until you come to think "can I even find 1 > million customers who understands how to use this thingy?", ... > > The HD manufacturer would probably in that situation just put on the > ethernet interface, additional RAM, verify that it can run some OS'es, > then make some OEM deals with a linux server company to design and support > the OS and utils that will go into it. ....
> One possible intermediary is to sell a small board that attaches to the > bottom of the drive, with a pass-through power connector and a bridge to > the IDE port. Would break the 3.5" form factor though.
What about a board that fits in a 5 1/4" bay, leaving room for a 3.5" drive below it, and has ARM, VLSI IDE controller, ethernet, ...
In a standard low-cost box designed for a single 5 1/4" device like CD-ROM or tape drive, this becomes a standalone network server. If you need a bunch, put them in a big case designed for a PC server or a RAID array.
The 5 1/4" form factor gives you enough room to have connectors for two IDE controllers, so you can use standard motherboard chips for the IDE and run two drives fast or four much less so. It also allows you to support fairly large memory for some applications.
This doesn't take a drive manufacturer. Anyone building PC motherboards or embedded systems could take a crack at it.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |