| 1 |
Matthias Bethke schrieb: |
| 2 |
> Hi Florian, |
| 3 |
> on Wed, Aug 27, 2008 at 10:29:07PM +0200, you wrote: |
| 4 |
>> Note1: NEVER EVER build some kind of RAID other than "Linear" (also called |
| 5 |
>> JBOD) over two IDE disks on the same cable. Performance will suffer greatly |
| 6 |
>> as will security because most simple onboard controllers can't handle a |
| 7 |
>> dying disk and that one might take the other one with it into death. |
| 8 |
> |
| 9 |
> Your suggestions sound reasonable (as reasonable as you get if one |
| 10 |
> insists on going with the drives that are there instead of getting a |
| 11 |
> third 500G drive that is :) [and for RAID5 I'd add a cheapish SATA |
| 12 |
> controller as well]) but I wonder why the above should be better than a |
| 13 |
> RAID0. The risk is the same---if either disk dies, the partition is |
| 14 |
> fuct. And considering drive mechanics are still the slowest part of the |
| 15 |
> system, even two EIDE disks that tend fight for the bus should be a tad |
| 16 |
> faster when striped than any one alone, which is what you effectively |
| 17 |
> get in a JBOD, right? |
| 18 |
> |
| 19 |
> cheers, |
| 20 |
> Matthias |
| 21 |
> |
| 22 |
|
| 23 |
Hmm, you might be right. Maybe someone should do a field test. |
Archives