Archived from groups: comp.sys.ibm.pc.hardware.storage (
More info?)
Previously WiderNetTech <techsupport@widernet.org> wrote:
> I work for a non-profit project at The University of Iowa and we want
> an inexpensive way to duplicate 250GB and larger hard drives containing
> content collected from the Internet in an off-line browsable and
> searchable format. These drives are then used in schools, clinics and
> universities in areas of the world where Internet access is extremely
> limited or non-existant.
> When I copy one drive to using SATA, I can copy 173GB in about 3 hours
> and 45 minutes (+/-900MB per minute). When our collection was smaller
> and we used IDE drives we would use a second-hand drive duplicator and
> we could duplicate drives at 450MB per minute. We have even tried using
> RAID1 arrays to duplicate drives, but the RAID tables persist and cause
> problems later on which result in corruption of the data.
> If we were able to copy drives at a rate of 1GB per minute, that would
> be great.
Should be doable with PCI and no more than, say, 4 drives total.
For more drives you may need PCI-X or PCI-E as bus and maybe
more than one controller.
> I will suggest to one of our student programmers the idea of writing a
> program to copy raw blocks from one device to several devices and see
> what we they can do.
It is not really difficult. There could be some system limits
you run into, but basically I think hacking, e.g., "dd_rescue"
(code available here: http://www.garloff.de/kurt/linux/ddrescue/)
to accept more than one target should do the job.
> Thanks for your suggestions and interest. I'll post back what we end up
> using for a solution.
Please do.
Arno