1 |
On Wed, June 26, 2013 00:13, William Kenworthy wrote: |
2 |
> On 26/06/13 04:59, Alan McKinnon wrote: |
3 |
>> On 25/06/2013 21:10, Mick wrote: |
4 |
>>> Hi All, |
5 |
>>> |
6 |
>>> I am considering my options for a new rig destined to last a few years |
7 |
>>> and one |
8 |
>>> of the Dell machines on offer has this Intel SRT fake-raid feature, |
9 |
>>> which |
10 |
>>> after some cursory googling, I am not entirely sure will work with |
11 |
>>> Linux. |
12 |
>> |
13 |
> |
14 |
> It will probably work quite well ... easy to set up, use and is reliable |
15 |
> ... until the motherboard fails and you find that ALL your data is now |
16 |
> inaccessible until you buy a compatible motherboard ... which may not |
17 |
> exist! - been there, done that, never again :) From memory there was no |
18 |
> linux driver needed (it was all done in the chipset). |
19 |
|
20 |
My experience with fake-raid was that a Linux-driver was needed. Maybe you |
21 |
had it in the kernel? |
22 |
|
23 |
> Use soft raid, its performance is at least as good (there was a report |
24 |
> saying it was usually better, even against some lower end dedicated raid |
25 |
> cards which were resource constrained), and its portable. |
26 |
|
27 |
Linux softraid outperforms any fakeraid card. The reason for this is simple: |
28 |
The drivers for softraid and the SATA/SAS/SCSI chipsets are still being |
29 |
improved by the kernel developers. |
30 |
|
31 |
The drivers for fakeraid are written on a friday afternoon and when they |
32 |
reach the stage of "it works", development stops. |
33 |
|
34 |
I've done some performance tests with fakeraid vs. linux softraid in the |
35 |
past. The difference in speed was shocking. |
36 |
|
37 |
> Thats not to say "dont use the box" - the disk interfaces are usually |
38 |
> very good performers in standard mode so just dont use the raid mode. |
39 |
|
40 |
Set it to "AHCI" mode for better performance on most new mainboards. |
41 |
|
42 |
-- |
43 |
Joost |