You are not logged in.
Pages: 1
I am getting another drive or two, either one 1TB or 2 500GB, and my other drive is coming back from RMA soon so I will have:
2 x 320gb
1 x 1TB
1 x 500gb
or
2 x 320gb
3 x 500gb
Should I just leave them as is or use RAID? I'm not familiar with RAID at all and am curious what benefits would be and how simple/difficult to set up ??
TIA ![]()
Offline
Basically, you give up some storage space for a minimal performance increase and redundancy, so if one disk fails, you can still run (but you still need backups).
Though the process described in the wiki looks complicated, it isn't too bad.
There really isn't a requirement that your disks be the same size, since the software raid can just take equally sized partitions. If you use lvm, either setup can be pretty versatile as far as managing your actual logical volumes (containing filesystems).
Either setup can work pretty well, though the 3x 500G might be marginally faster and cheaper and louder.
Offline
So it wouldn't really make much of a difference ??
Offline
So it wouldn't really make much of a difference ??
depends on how u set the RAID up.
Offline
2 x 320gb, 1 x 1TB, 1 x 500gb
Gives you: 4 partitions at 320Gb, 2 @ 180G, 1 @ 500 Gb,
or: 3 @ 320Gb, 2 @ 500, 1 @ 180G
2 x 320gb, 3 x 500gb
is: 5 @ 320G, 3 @ 180G, or just one size of disk in 2 arrays.
Doing this kind of breakdown shows that 5 disks is more flexible, but maybe the extra electricity requirement, and increased failure rate means that 4 disks would be better?
Note, it might be a good idea to put /boot on a small raid 1 array spread across all disks.
Last edited by vogt (2008-02-07 20:16:55)
Offline
Okay, I am going through the RAID Wiki on wikipedia and it makes some sense but to be honest this is so new for me, I can't really understand exactly what I could do with my drives and what the benefit would be.
Now let's say I have the two 320gb, one 1TB and one 500gb. Right now I have the 320gb partitioned so that one part is XP, one is Arch / and one part is /home, and the rest is a part for storage, as are the other 320gb, the 1TB and the 500gb.
So which RAID configuration would I use, and why? Also, would setting RAID up require me to wipe my drives and kind of "start over," or do I just leave the drives as they are and then set up RAID in my mobo BIOS?
Would there be anything I have to setup/configure in Arch if I change to RAID?
Sorry for so many questions lol, I'm just super curious and interested in this
Thanks
Offline
Thanks vogt (and hello fellow Torontonian
) that clears a bit up for me, though my questions still stand.
Let me get this straight: RAID basically allow me to take all my drives as one big one and then "redo" them however I choose?
Offline
Yes, however more than that it allows you to either sacrifice diskspace for redundancy (1) or sacrifice security for speed (0).
I believe LVM is capable of making all your disks into one entity, Just a Bunch Of Discs. aka Concatenating.
There is also the differences between fully softwaremanaged RAID, Bios/software managed (FAKE) and True Hardware RAID.
Some manufacturers of hardware RAID controllers have beautiful flash-animations of different levels of RAID.
Striping and mirroring are the two most used modes. (RAID0 and RAID1). Mixed versions require checksums to be calculated and this draws some cpu-time, not much thou. (mostly raid5)
Edit: FAKE raid (BIOS/driver) is alot easier to set up and somehow feels more pure, but it's not as flexible as software. If you are considering some way of dualbooting then I'd say go with fake-raid. Performance-wise it really doesn't matter.
Yes, Fakeraid will wipe your disks.
Software based let's you choose what partitions on the disks to use for raid sets so you can leave some part of the disk untouched.
Last edited by loosec (2008-02-07 21:18:11)
Offline
Ah, windows is pretty inflexible about moving around and still being able to boot, so if you still need it, you'd have to reinstall it.
Unless you need to access the arrays from windows, don't use the raid provided by your bios, as it's a bit inflexible (it is then hardware dependent), and offers no advantages. More here.
It is possible to convert between raid levels, adding disks to a [sul=http://wiki.archlinux.org/index.php/Convert_a_single_drive_system_to_RAID]raid1[/url], or to raid 5 etc. with this unmaintained tool.
So you are probably better off making a backup, then trying to convert to your desired configuration via raidreconf, expecting it to destroy your data; while if you are going to use lvm in addition (recommended), you basically have to wipe the disk.
As stated in the wiki, there are a couple of thing to do:
change your mkinitcpio.conf to include something like:
HOOKS="base udev autodetect pata scsi sata raid lvm2 uresume filesystems"
md=0,/dev/sda5,/dev/sdb2run
# # mdadm -D --scan >>/etc/mdadm.confand add your array(s) that are part of the volume group containing your root logical volume to the kernel parameters in /boot/grub/menu.lst
kernel /vmlinuz root=/dev/mapper/vg0-root md=0,/dev/sda2,/dev/sdb2,/dev/sdc2...And that is after you set up the array using mdadm (which happens to have a helpful manpage)
2 x 320gb, 1 x 1TB, 1 x 500gb
Gives you: 4 partitions at 320Gb, 2 @ 180G, 1 @ 500 Gb,
or: 3 @ 320Gb, 2 @ 500, 1 @ 180G
As far as the types of arrays, it depends on the level of redundancy vs performance you want (also the space you want to leave windows).
Low redundancy: 4 x 320G raid5, 2@ 180G raid1, 1x 500G is then windows and/or (partly) just added to your volume group, for disposable stuff like /tmp, or your pacman cache. You could go raid0, but loosing everything when one of four disks crashes is a bit too risky in the long term.
Last edited by vogt (2008-02-07 21:30:21)
Offline
Okay, it's making more sense now.
My ideal scenario as far as usage is: 1 partition for XP, 1 for Arch /, 1 for /home and of course need 1 swap, so that's 4 partitions, and they can all be small-medium in size. The rest of space would be for generic things like mp3, videos, documents, etc so no specific sizes or anything are necessary for me.
Would it make sense to do:
1 x 100gb for XP
1 x 20gb for /
1 x 100gb for /home
1 x 4gb for swap
Then the rest can be split up however ??
Offline
Use lvm on top of your raid: it makes it easier to combine all your storage, so none is unusable, while you can change the sizes easily. (only your root would have to be resized from a livecd), as well as move the logical volumes to different physical volumes online.
Also worth trying would be level 10 which happens to work for >1 disk.
Offline
Okay this is sounding like a good idea, and I am going to be reinstalling XP anyway this weekend. However, how much of a speed increase really is there, and how much more secure is my data with RAID than just using my drives as are??
Offline
I use to run software raid on two 250 gb drives. The reason I did was because that's the way windows was set up on the computer when I bought it. It wasn't too difficult to set up. The wiki was a huge help. Much thanks to the person who wrote it! Later, once I was more familiar with raid in general and my hardware / (nv bios) I decided to get rid of the raid and use the disks independantly. For a backup, I still keep a clone of my main arch install on the secondary disk (When ever I feel like my system is running really solid, I'll reclone my main install, or just boot into the clone and update it if I hadn't made too many changes to the main installl.) For me, this was much more practical, and the exrta disk space gained comes in handy, and if one of my disks fail, I still have a system I can boot into.
-- archlinux 是一个极好的 linux。
Offline
Pages: 1