On Thu, Nov 14, 2013 at 5:20 PM, m.roth@5-cent.us wrote:
James B. Byrne wrote:
On Thu, November 14, 2013 12:51, Reindl Harald wrote:
Am 14.11.2013 18:23, schrieb James B. Byrne:
From what I have read it appears that the system disk must use RAID 1 if it uses RAID at all.
So, this is saying, if I read it aright, that one can have multiple RAID arrays spread over the same spindles but each in differing partitions. Is that right?
<snip>
In principle it is ok. Issue here is making sure you will not loose stuff if a drive takes a dump. It is easier to have different drives for different raids because you then can limit the damage. At leas that is what I was taught in kitty school.
BTW, I intend to install CentOS-6.4 with software RAID as the eight disks are mounted in the system chassis. As far as I can tell, there is no
hardware
IMHO software raid is not bad in principle, and the Linux one (mdadm) supposedly is rather decent. Think this way: you can grab those drives, slap them into another linux box, and rebuild the array. I could swear i head a presentation from a Dell engineer they thought the hardware raid was too slow. Also, this is the proper time to mention ZFS.
RAID controller (unless there is one on the MB, in which case SW Raid is likely a better choice anyway.
Big question: what manufacturer, and what support chips? Most of our Dells have PERC 6xx or 7xx, which is good hardware RAID. We have several boxes with the Intel RAID on chip - aka fakeRAID. They are a *pain*, and we had them just present unRAIDed drives, and used the very nice to use Linux software RAID.
Isn't the PERC a rebadged LSI?
mark
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos