Raid (5) Question
Once a raid is created it is seen as a block device.
So, can a md device be paritioned like any other block device and is it still limited to a max of 4 primary paritiions? Would one use extended partitions to add more? Thanks, WT |
Short answer: yes and yes. You can treat the lun like a normal single drive. =D
|
Thanks for your reply.
Here's the situation I was trying to work out in my head. For a regular drive. The above scenario is not a problem. However, if I grow my raid (5) this makes things a little trickier, especially if I did not anticipate this. For example, if I start with a 2TB raid 5 and partition it into 4 x 500 GB partitions all is well. What if I needed to grow this partition by another TB? It looks like I'm out of luck. Am I correct, or is there something I'm not taking into consideration? Thanks, WT |
Depends on what your using. If your using the plain vanilla software raid, then you're SOL if you wanna do anything with it. If you're using any sort of virtualized storage layer (like LVM, VSF, or ZFS), then you can grow it as you need by simply adding drives to the system and using the appropriate command. However, if I may ask, why use RAID5? RAID10 will get you much better performance at a not-as-bad-as-it-seems loss of drive space. RAID5 is really only acceptable for read-heavy, write-light operations IMHO.
|
Quote:
Yes, you are going exactly where I want you to: LVM. I really have never used it before. My partitions for fixed drives have been fine. Now that I want to set up this raid and make it flexible for future use, it seems like I have to take LVM as well in order to get the max benefit. I had heard that LVM can make data recovery a pain, but this may have been geared towards / being LVM. Are there any down sides to LVM for a NAS raid? If I set up LVM I am probably going to be hosed anyway since I was going to use JFS. JFS can only grow its FS. So I don't think it would be useful for LVM if I wanted to decrease a LV. Correct? I didn't want to use XFS since it may lose data in the event of power loss, but the raid 5 write hole is a much bigger risk (I beleive) and am thus getting a nice beefy UPS to email me (I travel for work) and shutdown gracefully. Any thoughts? Thanks, WT |
DISCLAIMER: I found ZFS early on and haven't looked back since, so I'm not quite up to date on other virtualized storage layer technologies. I hear btrfs is supposedly to be an end-all?
I think if you want to be as safe as possible, I'd take a bit of a capacity hit for mirrored storage on a box running ZFS. Set it to scrub maybe weekly, and call it a day. Please glance at the following link for a (tiny little bit biased, but ultimately true) comparison of RAID levels: http://www.miracleas.com/BAARF/RAID5_versus_RAID10.txt A quote from the above link: If a drive costs $1000US (and most are farhttp://blogs.sun.com/roch/entry/when_to_and_not_to is also a great read. As far as increasing and decreasing logical volumes go, it's always a lot harder to reduce capacity than it is the increase it. For home use, I would recommend hosting everything (backups, share folders) on one behemoth file share, then backup your backups to tape. I would keep your VM's in a separate pool, depending on performance issues, but if it's for home, it'd likely be fine in the same box. UPS = always a good idea. |
Quote:
What are you running for a distro? WT |
Personally, I use Debian for applications (web, PXE, etc), OpenSolaris for FS, Windows for AD, and Windows and Ubuntu for workstations. I haven't taken the time to learn BSD, so I couldn't speak on it's merits. But use what's best for you...
I'm not too sure about BtrFS, all I know is that it's new and it's still in development, probably best to steer clear for now, unless you've got a fair bit of experience and some storage to play around with. |
All times are GMT -5. The time now is 07:08 AM. |