On Sat, 2005-06-18 at 16:58 +0200, Dag Wieers wrote:
Hi, On my x86_64 system I have a SiL311x controller that can do RAID. If I configure my 2 identical disks in a RAID1 setup, I would expect to see only 1 block device on Linux. Still I see 2 block devices. Is this intentional, and if so, isn't that dangerous ? (i.e. writing to both disks at the same time) Anyone with an insight, please explain :)
They are _not_ RAID, they are what I call "Fake/Free RAID" (FRAID).
The BIOS just allows you to designate the disks as RAID-1 (simple mirroring) or organize the blocks into RAID-0 (stripped). Once the OS loads, it not only needs to know that so it doesn't see the disks as "just a bunch of disks," but so it knows how to write the data as such. That is done 100% in the OS driver.
Since that logic is typically licensed from the same 2-3 3rd parties, and those 3rd parties make royalties on those sales, it is _never_ going to be GPL.
Now there is a "clean-room" GPL RAID logic in "ataraid.c" in the kernel. And then there are "card interface" sister drivers in "hptraid.c" (HPT), "pdcraid.c" (Promise) and "silraid.c" (Silicon Image), but they _rarely_ work well. Little changes and differences in the cards, drivers, etc... always seem to cause all sorts of issues when I've tried them.
_Real_ hardware RAID cards _never_ talk directly to the ATA channels. They _always_ have _all_ communication between the system and the ATA channels go through an "on-board intelligence" like an i960 microcontroller in the Promise SuperTrak, Adaptec 2400/2800A, LSI Logic MegaRAID 320-4 and 320-6, an XScale superscalar microcontroller in the LSI Logic MegaRAID 320-8X, or an ASIC in the 3Ware Escalade and NetCell SR-series.
For more on ATA RAID, see the article "Dissecting ATA RAID Options" in 2004 April "Sys Admin" magazine. The article is not available on-line, but if you subscribe to Sys Admin, you get a CD with _all_ back issues of Sys Admin and Perl Journal: http://www.samag.com/articles/2004/0404/