Boston Linux & Unix (BLU) Home | Calendar | Mail Lists | List Archives | Desktop SIG | Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings
Linux Cafe | Meeting Notes | Blog | Linux Links | Bling | About BLU

BLU Discuss list archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Moving from RAID 0 to LVM RAID?

 [If you're going to cross post your questions between BBLISA and BLU, 
then I guess I'll cross post my answer as well.] 

Scott R. Ehrlich wrote: 
> Now, from the OS side, LVM is an option.  Say...I want to use more 
> than two disks, and I don't want the RAID 0 problem again. 

Can you use LVM? Is that the implied question? 

LVM and software RAID are related, but different. LVM doesn't provide 
redundancy. Its primary purpose is combining multiple block devices and 
presenting them as one block device, where you can then put a file 
system. A typical setup will layer LVM on top of RAID, permitting the 
addition of other physical disks or another RAID set to the LVM volume, 
but you can use LVM directly with raw disks. 

So the answer to the above question is no. Using LVM (without RAID > 1) 
to spam multiple disks will lead you to the exact same problem as RAID0. 
(It's not quite as bad, because the all data isn't evenly distributed in 
chunks across the disks. I've read it is possible to recover files from 
the surviving disks in a failed LVM set.) 

> When I get a replacement disk and build the system from the ground up 
> again, I could, conceivably, use hardware RAID 1 for the OS on two 
> disks, and CentOS 5 64-bit's LVM for software RAID 5 (or maybe 1+0 if 
> available) on the remaining for 4 disks, maybe 3 disks as active and 
> the 4th as a hot spare? 

The rule of thumb is use RAID 1, unless you can't afford it, then RAID 
5. RAID 6, a variation on RAID 5, is becoming popular these days, which 
is a step more fault tolerant. 

> I've never had much faith in software raid, since it is not 
> hardware-based, and there would be a performance hit, but in this 
> case, it could be an option. 

Modern software RAID implementations are well proven to be reliable and 
in wide use. Performance on typical low-end servers is often better 
using software RAID than entry-level hardware RAID. Modern day CPUs 
don't sweat the parity calculations. A bit of research online should 
turn up evidence of this. 


Tom Metro 
Venture Logic, Newton, MA, USA 
"Enterprise solutions through open source." 
Professional Profile:

This message has been scanned for viruses and 
dangerous content by MailScanner, and is 
believed to be clean. 

Discuss mailing list 
[hidden email]

BLU is a member of BostonUserGroups
BLU is a member of BostonUserGroups
We also thank MIT for the use of their facilities.

Valid HTML 4.01! Valid CSS!

Boston Linux & Unix /