POLL: Which RAID level do you use?

Which RAID level do you use at home?

  • RAID 0

    Votes: 14 23.0%

  • RAID 1

    Votes: 5 8.2%

  • RAID 10

    Votes: 3 4.9%

  • RAID 0+1

    Votes: 2 3.3%

  • RAID 3

    Votes: 0 0.0%

  • RAID 4

    Votes: 0 0.0%

  • RAID 5

    Votes: 20 32.8%

  • RAID 6

    Votes: 17 27.9%
  • Total voters
    61
H

holzmann

Weaksauce
Joined
Jan 20, 2011
Messages
84
Please vote and then explain why.

Also, do you use hardware (dedicated controller) or software RAID?

Thanks.
 
1, 10, 4, 5 & 6... Not sure which one to choose to vote with?
 
RAID10. Went for reliability and speed at the expense of storage efficiency. Thought about RAID6, but then went this route for no particular reason in the end.
 
Thecus N5200BR NAS
RAID 5 Array (software)
5x Seagate 750GB (ST3750640AS) HDs

RAID5 seems to offer the best balance btw performance and data safety.
 
RAID0 FTMFW. Go big or Go home. Speed Matters. Fuck Reliability. That's what cheap 2TB WD Caviar drives are for.
 
I use RAID 5.

I have a Synology NAS (DS1010+) which uses software linux MD RAID. I choose 5 because I only have 5 drives and wanted 8TB instead of only 6 and 1-drive parity is reasonable redundancy for 5 drives.
 
At work either raid 5 or 6 on all non boot arrays. Now 40 to 50 TB of linux software raid in 12 to 15 arrays. On boot arrays raid 1 across all disks meaning I do have 8 drive raid 1 arrays that all 8 members are mirrors of each other.

At home I removed all raid because it uses more power and prevents my htpc machines from spinning down disks to save power. In this case I use individual 2TB green drives and the htpc software takes care of managing and balancing the usage between all drives in the media store.
 
Last edited:
RAID 0 on two Spinpoints. I have an extrenal with all my media, and all documents on dropbox so i don't care of a failure (though haven't had one in the 2 years I've been on RAID 0). Speed difference is significant.
 
Main Computer: RAID 0, Hardware Mobo/"Fake" Raid, OS Drive
Storage Server: WHS, no raid level. :(

-Cool-
 
notarat said:
RAID0 FTMFW. Go big or Go home. Speed Matters. Fuck Reliability. That's what cheap 2TB WD Caviar drives are for.
Click to expand...

RAID 10/0+1 is capable of being just as big as RAID 0, done right is faster than RAID 0, and has redundancy included.

Technically RAID 0 isn't even a RAID.
 
Currently using RAID 5 for a 5-disk 1.2TB volume, storing digital media. Viable RAID level for the disks I had to hand.


Tolyngee said:
RAID 10/0+1 is capable of being just as big as RAID 0, done right is faster than RAID 0, and has redundancy included.
Click to expand...

Even the Linux RAID 10 driver, which is a better implementation than most other RAID 10 drivers (it can read from all disks, rather than only half), isn't faster than RAID 0. It can match RAID 0 read speeds, but writes will suffer. Usually though, RAID 10 is a fair balance between (write) speed and reliability. Cost of implementation is higher than RAID 5 or 6, though.
 
Last edited:
RAID 0 and RAID 1 are both simple and straightforward, but I am still trying to get my head around RAID 5. It's a combination of both? Can someone please explain?

I am looking for a home backup solution that offers its own redundancy. I currently run RAID 1 with two 500GB drives with FreeNAS and a 3ware PCI controller.

I want to upgrade to 2TB of redundant storage.

I like FreeNAS, so I will probably stick with it.

Do I still with the 3ware controller or go with software RAID?
Stick with RAID 1 or go to RAID 5?
Get WD drives with TLER or not? Are they the most reliable?
 
For freenas you would get better reliability and redundancy by ditching the hardware raid card and using zfs and one of its redundancy levels.
 
Linux softwareraid 6. Using level 6 for redundancy and awsomeness.
In the end I will use 20 disks.
 
holzmann said:
RAID 0 and RAID 1 are both simple and straightforward, but I am still trying to get my head around RAID 5. It's a combination of both? Can someone please explain?
Click to expand...

RAID 5 is like RAID 0, but with data reconstruction. Unlike RAID 1, which simply replicates data between two or more disks, RAID 5 (and its cousins RAID 4 and RAID 6) calculates reconstruction data, and writes that to the array in addition to the user's data.

Since RAID 5 stripes data among the drives in the array in the same way that RAID 0 does, if a disk goes missing, you have one drive's worth of "holes" in all of the data on the array. Unlike RAID 0 though, RAID 5 uses that reconstruction data to calculate what that missing data should be, thereby maintaining data integrity.

drescherjm said:
At home I removed all raid because it uses more power and prevents my htpc machines from spinning down disks to save power. In this case I use individual 2TB green drives and the htpc software takes care of managing and balancing the usage between all drives in the media store.
Click to expand...

I certainly understand the power saving angle, but what do you do if one of those drives fails? Would you consider something like FlexRAID?
 
Software RAID5 and RAID1.

OP, don't forget, software RAID does not equal FlexRAID, they may both use the CPU but both are very different.

As for TLER, it is only necessary with certain 3Ware controller cards and low-end hardware RAID cards that can't handle spin-up/down very well. All it does is add on time to the drive so the RAID controller doesn't freak out thinking the disk is missing or broken. It isn't necessary with software or FlexRAID.
 
Red Falcon said:
As for TLER, it is only necessary with certain 3Ware controller cards and low-end hardware RAID cards that can't handle spin-up/down very well. All it does is add on time to the drive so the RAID controller doesn't freak out thinking the disk is missing or broken. It isn't necessary with software or FlexRAID.
Click to expand...

Actually the purpose of TLER is in the meaning of its acronym: Time Limited ERROR Recovery. If the drive hits a bad sector, normally it will try and retry to write data to that sector - eventually it will give up, but this can take up to 30 seconds, during which time the drive isn't talking to the RAID controller; the RAID controller eventually gives up on the drive and marks it as failed.

TLER prevents this by limiting the time spent by the drive in error recovery mode, to around 7 seconds. After this time has elapsed, the drive tells the RAID controller that the write has failed, and then it is up to the RAID controller to deal with the situation.

Look here for more information.
 
parityboy said:
RAID 5 is like RAID 0, but with data reconstruction. Unlike RAID 1, which simply replicates data between two or more disks, RAID 5 (and its cousins RAID 4 and RAID 6) calculates reconstruction data, and writes that to the array in addition to the user's data.

Since RAID 5 stripes data among the drives in the array in the same way that RAID 0 does, if a disk goes missing, you have one drive's worth of "holes" in all of the data on the array. Unlike RAID 0 though, RAID 5 uses that reconstruction data to calculate what that missing data should be, thereby maintaining data integrity.



I certainly understand the power saving angle, but what do you do if one of those drives fails? Would you consider something like FlexRAID?
Click to expand...

Thanks for the great explanation.

I still think that RAID 1 might be the solution for me...
 
I run a zfs zraid2... Its kind of like a RAID-6 (double parity). I run it on my FreeBSD file server for storing backups and multimedia.
 
parityboy said:
I certainly understand the power saving angle, but what do you do if one of those drives fails? Would you consider something like FlexRAID?
Click to expand...

My HTPC backend currently runs on 2 linux machines running mythtv since May 2004.

Code: 
Number of shows:
195
Number of episodes:
3108
First recording:
Sunday May 30th, 2004
Last recording:
Tuesday February 1st, 2011
Total Running Time:
6 years 8 months 4 days 11 mins
Total Recorded:
4 months 27 days 21 hrs 45 mins
Percent of time spent recording:
1%

I backup 200 to 400 GB of the most important data (os + user data + programming + mythtv database ...) I have on these machines however the video I do not currently backup. Most of it I could just reschedule.

Although with the price of 2TB drives now I may eventually buy some more to backup the video as well I would probably just add an additional backup job for this and put the video in its own backup pool.

Edit: Outside of that I do monitor the smart status of my drives. At one point I had nagios monitoring the smart for each drive (along with other things) sending me email updates on changes but my ISP broke the email sending and I have not spent time to fix that.
 
Last edited:
drescherjm said:
For freenas you would get better reliability and redundancy by ditching the hardware raid card and using zfs and one of its redundancy levels.
Click to expand...

Thats what I thought until the box lost power & I had a corrupted pool metadata. No way to reimport the pool on opensolaris, freenas, or FreeBSD 8.

Freenas uses a VERY old version of zfs. Do a real freebsd box instead, IMHO.
 
You must log in or register to reply here.
Back
Top