| 1 |
Patrick Lauer wrote: |
| 2 |
|
| 3 |
>On Wed, 2006-02-01 at 07:54 -0500, Sean Cook wrote: |
| 4 |
> |
| 5 |
> |
| 6 |
>>-----BEGIN PGP SIGNED MESSAGE----- |
| 7 |
>>Hash: SHA1 |
| 8 |
>> |
| 9 |
>> |
| 10 |
>> |
| 11 |
>>>They are all different questions: |
| 12 |
>>> |
| 13 |
>>>* Hardware RAID vs Software RAID |
| 14 |
>>>Hardware RAID offers more performance and often more sophisticated RAID |
| 15 |
>>>features. |
| 16 |
>>> |
| 17 |
>>> |
| 18 |
>That is debatable :-) |
| 19 |
>For really high-end systems hardware raid (3ware, SAN box, ...) is more |
| 20 |
>performant, but even on a modestly loaded Athlon64 you'll (usually) be |
| 21 |
>I/O limited before CPU limits become visible |
| 22 |
> |
| 23 |
> |
| 24 |
It is, nearly everything surrounding performance on RAID is debatable. |
| 25 |
Depends on your usage and controller and disk-types etc. |
| 26 |
Dependant on what you do you might hit an I/O limit faster than an |
| 27 |
CPU-limit. |
| 28 |
Agreed. |
| 29 |
|
| 30 |
>>>Another advantage is that Hardware RAID controllers often offer you the |
| 31 |
>>>opportunity to extend your RAID array beyond the usual 4 SATA |
| 32 |
>>>interfaces. Depends on the card though. |
| 33 |
>>> |
| 34 |
>>> |
| 35 |
>I'm running a software raid spanned over 3 controllers ... I'm not aware |
| 36 |
>of a hardware-based solution that even comes close to that flexibility. |
| 37 |
> |
| 38 |
> |
| 39 |
How many disks can you put in ? |
| 40 |
AFAIK most motherboard interfaces support 4 SATA disks |
| 41 |
|
| 42 |
>>Back in the day... we were running the LSI MegaRAID controllers with |
| 43 |
>>RAID 1 on 9gig SCSI disks, we were choking our application |
| 44 |
>>(mysql/mod_perl) we took the MegaRAID controller out and performance |
| 45 |
>>increased by about 25%. (note: this is a real world situation) |
| 46 |
>> |
| 47 |
>> |
| 48 |
>Even if this is anecdotal - my experience is very similar. |
| 49 |
>Software raid5 on an Athlon-1000 (basically a box of leftovers) takes |
| 50 |
>~7% CPU to saturate a 100Mbit line. An older 3ware controller maxed out |
| 51 |
>at similar speeds, slightly lower CPU load - but adding more RAM and a |
| 52 |
>slightly faster CPU is in this case much cheaper and faster than buying |
| 53 |
>a new controller. Also changing controllers might not be possible |
| 54 |
>without wiping and rebuilding the array. |
| 55 |
> |
| 56 |
>But if you need a disk array with maximum performance I'd still suggest |
| 57 |
>a hardware-based solution. |
| 58 |
> |
| 59 |
> |
| 60 |
Me too :-) |
| 61 |
But considering your comments, why ? |
| 62 |
|
| 63 |
Grtz Ramon |
| 64 |
|
| 65 |
-- |
| 66 |
To be stupid and selfish and to have good health are the three requirements for happiness, though if stupidity is lacking, the others are useless. |
| 67 |
|
| 68 |
Gustave Flaubert |
| 69 |
|
| 70 |
|
| 71 |
|
| 72 |
|
| 73 |
-- |
| 74 |
gentoo-server@g.o mailing list |
Archives