Anyone using large external RAID arrays?

[VirtualRain]

I'm curious who's using large external RAID arrays (12+TB) of HDs with their Mini?

What enclosure are you using?

What RAID level are you using? RAID5, 6, 10, ???

Have you ever had to rebuild the array? Has that been problematic?

Of course, the common wisdom is that RAID5 is dead (too risky with large arrays) but I'm curious if anyone's defying this in practise.

In my case, my media collection has out grown my current collection of JBOD drives and it would be a lot of effort to rebuild if I lost a disk, so I'm going to implement an external RAID array for my Mac Mini Plex media center and I think I'm pretty much settled on RAID10 due to the issues with parity RAID on large arrays, but interested in other's experiences.

 

[ColdCase]

What is your backup strategy? Many use OSX to RAID0 a bunch of JBOD drives (three 4TB drives for 12TB) and then use a backup strategy (TM or clone to second RAID0 drive set perhaps). Not as automated as RAID10, but there is less chance the system corrupting all your data.

Since there is no parity, recovery of RAID10 is about as quick as duplicating data (Read and right) similar to a main drive array and its backup. When something goes wrong with the main array, fix it and copt the data back to it from the backup.

RAID 10 systems will automate the process and perhaps has a cleaner interface, but at a price.

 

[VirtualRain]

What is your backup strategy? Many use OSX to RAID0 a bunch of JBOD drives (three 4TB drives for 12TB) and then use a backup strategy (TM or clone to second RAID0 drive set perhaps). Not as automated as RAID10, but there is less chance the system corrupting all your data.

Since there is no parity, recovery of RAID10 is about as quick as duplicating data (Read and right) similar to a main drive array and its backup. When something goes wrong with the main array, fix it and copt the data back to it from the backup.

RAID 10 systems will automate the process and perhaps has a cleaner interface, but at a price.

Thanks for your reply. I honestly don't see any advantage to RAID0 with a backup vs RAID10 especially for a media library where top performance is not a critical requirement. In fact, with RAID0 and a backup, a drive failure will take you offline and require rebuilding the entire array. A drive failure in RAID10 only requires rebuilding a single drive and you have full access to your library during the rebuild. I think data corruption is possible (although unlikely) with either scheme and since this is a media library, data integrity is not my top priority, saving time rebuilding the library in the event of a drive failure is more of a concern.

 

[drsox]

Thanks for your reply. I honestly don't see any advantage to RAID0 with a backup vs RAID10 especially for a media library where top performance is not a critical requirement. In fact, with RAID0 and a backup, a drive failure will take you offline and require rebuilding the entire array. A drive failure in RAID10 only requires rebuilding a single drive and you have full access to your library during the rebuild. I think data corruption is possible (although unlikely) with either scheme and since this is a media library, data integrity is not my top priority, saving time rebuilding the library in the event of a drive failure is more of a concern.

I have 11 TB in a RAID5 over 5 drives. I have a second identical unit that I mirror to once a week. I don't generate much content over a week so I'm only using the RAID to guard against a single drive failure (although I could modify it to be dual drive resilient if I added a 6th drive). Having 2 identical units means I'm also protected against hardware failure (can just move the RAID set from one unit to the other).

RAID rebuild from a single drive change is easy and takes 6 hours or so. I've also just reset the mirror unit and rebuilt it from scratch - took 3 days in total, most of which was pouring the data back.

I've seen (well heard actually) some data corruption but if I changed the units to newer ones that used BTRFS I could also apparently recover from a single bit error - see this article : http://arstechnica.com/information-...-and-atomic-cows-inside-next-gen-filesystems/

 

[FireWire2]

I'm curious who's using large external RAID arrays (12+TB) of HDs with their Mini?

What enclosure are you using?

What RAID level are you using? RAID5, 6, 10, ???

Have you ever had to rebuild the array? Has that been problematic?

Of course, the common wisdom is that RAID5 is dead (too risky with large arrays) but I'm curious if anyone's defying this in practise.

In my case, my media collection has out grown my current collection of JBOD drives and it would be a lot of effort to rebuild if I lost a disk, so I'm going to implement an external RAID array for my Mac Mini Plex media center and I think I'm pretty much settled on RAID10 due to the issues with parity RAID on large arrays, but interested in other's experiences.

I have 12TB (15TB RAW) storage holding my media content runs 24/7 for over 2 year. Upgraded from 8TB (5x 2TB Hitachi drives
http://www.amazon.com/tray-less-hardware-enclosure-eSATA-USB3-0/dp/B00E0T1BSW

 

[VirtualRain]

I have 12TB (15TB RAW) storage holding my media content runs 24/7 for over 2 year. Upgraded from 8TB (5x 2TB Hitachi drives
http://www.amazon.com/tray-less-hardware-enclosure-eSATA-USB3-0/dp/B00E0T1BSW

RAID5? 4x3TB? Ever have s drive failure? If so, what was your rebuild experience?

 

[now i see it]

Current spinning HD reliability now-a-days sucks big time as the capacity has increased over the years. Higher capacity = lower reliability. They all fail eventually.
The only answer I have found that gives you half a chance to keep your data is:
DROBO

 

[VirtualRain]

Current spinning HD reliability now-a-days sucks big time as the capacity has increased over the years. Higher capacity = lower reliability. They all fail eventually.
The only answer I have found that gives you half a chance to keep your data is:
DROBO

I don't think drive reliability is any worse or better now than it ever was. Mechanical drives will wear out, there's no doubt about it. However, the problem with larger drives is that they can cause problems with RAID5 rebuilds which is why I'm inquiring about real-world experiences.

I can see the appeal of Drobo but I'm not a big fan of their proprietary RAID technology. It just adds unnecessary complexity and attempts to hide it from the user. When it comes to large storage volumes, I'm all for simplicity.

While Drobo might give you half a chance to keep your data, a backup will give you a full chance.

I've opted to run my media library as a RAID0 array and backup to a set of JBOD drives.

 

[FireWire2]

RAID5? 4x3TB? Ever have s drive failure? If so, what was your rebuild experience?

Yes In five year using this RAID box... i did replace 1x 2TB - 1st 3yrs
None in 3TB yet . but I did have an incident the RAID's alarm went off.
While my raid was running. I did :
- Remove the issue HDD, by simply open the door
- Wait for 15 sec , re-inserted
- Within 45sec, the system rebuild.
It took about 18~20hrs

BTW it's not a 4x 3TB, it's 5x 3TB RAID5 to get 12TB . I can set up as 9TB RAID5+hot spare it my data is critical, I backup my picture only, but not other media content.

I'm thinking of going to use 6TB HDD, which I confirm with DATOptic, their RAID engine supports 6TB HDD

On Drobo is good in its own class, if it crashed, 99% will you lost your data, there is NO recovery data utility, that can recover data from DROBO

 

[VirtualRain]

Yes In five year using this RAID box... i did replace 1x 2TB - 1st 3yrs
None in 3TB yet . but I did have an incident the RAID's alarm went off.
While my raid was running. I did :
- Remove the issue HDD, by simply open the door
- Wait for 15 sec , re-inserted
- Within 45sec, the system rebuild.
It took about 18~20hrs

BTW it's not a 4x 3TB, it's 5x 3TB RAID5 to get 12TB . I can set up as 9TB RAID5+hot spare it my data is critical, I backup my picture only, but not other media content.

I'm thinking of going to use 6TB HDD, which I confirm with DATOptic, their RAID engine supports 6TB HDD

On Drobo is good in its own class, if it crashed, 99% will you lost your data, there is NO recovery data utility, that can recover data from DROBO

Thanks for sharing. This goes counter to popular thinking that RAID5 died with the advent of 3TB drives. In theory, at least, the chances of rebuilding an array that large without an unrecoverable error are pretty high. Either the risk is overstated or you're lucky. I have to think that unless you're going with enterprise grade drives, a RAID5 array built on 6TB drives is very unlikely to rebuild in the event of a drive failure. Personally, I wouldn't do it. You're probably much better off running some 6TB drives in RAID0 in a new enclosure and using your existing setup to back that up.