[Atlplasma]

My iMac died the other day. When I tried to recover on to the new iMac from Time Machine, I discovered that the fairly new 2TB drive was a brick. A few days later I went downstairs to play some music and was met with a blinking red light on my NAS. This 1TB drive was also fairly new. The tech at the Mac store said that hard drives have been much less reliable since the factories were flooded a few years ago. He noted that most manufacturers now offer only 1-year warranties.

So is there a best brand of drive? I know Hitachi used to have good reputation. How are you protecting your stored music? Are you using NAS in redundant configurations? Throwing out old drives after a year? Keeping backups on the cloud? Putting everything back on tape? 

[rpf]

I keep more than one backup for my music. Perhaps I should for my computer data as well.

For music, I've been using LaCie drives, which have performed well to date.

[Doublej]

What brand are you using? I'll stay away from them. I have had good luck so far with Western Digital. I run health checks on them regularly. RAID will eliminate problems of single dsik failure but not of a location based failure such as a house fire. IMHO cloud backup is best for piece of mind.

[trackball02]

I just had a Western Digital 2TB external drive go bad after using it only 3 times as a off site backup. After getting on the phone with WD support, they admit that is was also a bad drive, and I'm getting an exchange under warranty. 

I'm going to get another drive (non Western Digital) to have a back up of my backup in case the replacement WD fails. I'm not going to buy Western Digital drives anymore.

I'm using 2  3TB Segate Barracuda drives with my Synology NAS, that seem to be reliable.

[Atlplasma]

What brand are you using? I'll stay away from them. I have had good luck so far with Western Digital. I run health checks on them regularly. RAID will eliminate problems of single dsik failure but not of a location based failure such as a house fire. IMHO cloud backup is best for piece of mind.

All my external drives, including the ones that failed, are Western Digital. I have my music storage in a Raid enclosure and will probably buy one for my office iMac as well.

[Atlplasma]

I just had a Western Digital 2TB external drive go bad after using it only 3 times as a off site backup. After getting on the phone with WD support, they admit that is was also a bad drive, and I'm getting an exchange under warranty. 

I'm going to get another drive (non Western Digital) to have a back up of my backup in case the replacement WD fails. I'm not going to buy Western Digital drives anymore.

I'm using 2  3TB Segate Barracuda drives with my Synology NAS, that seem to be reliable.

I need to check my records on the WD drives. I think the 2TB is probably still under warranty. I've used Seagate in the past and had good luck with.

[Doublej]

Looks like the both Seagate and Western Digtial have issues now.

https://www.backblaze.com/blog/hard-drive-reliability-update-september-2014/


Hitachi is now owned by Western Digital so who know what will happen to those drives from a reliability perspective.

More reason to use RAID and/or the cloud for backup.

It you have multiple machines with free space on the hard drives you can use Crash Plan to back them up to each other.

[Wayner]

Large storage devices (1, 2 or more TBites) have multiple discs as well as multiple readers and suspensions. This all increases the odds of failure. While it is convenient to have large storage devices, the failure risk does increase because of the very reason I mentioned. The more dingle-berries, the higher the odds of failure.

One solution: smaller storage devices (1T or less) or several backups......

[dB Cooper]

Another option for music archiving might be blu-ray discs which can hold 50GB so even a large lossless collection would fit on a relatively moderate number of discs. Keep in a safe place (preferably offsite) and handle ONLY when needed, like after a catastrophe.

[JimJ]

I keep more than one backup for my music. Perhaps I should for my computer data as well.

For music, I've been using LaCie drives, which have performed well to date.

Off-site backups for everything have saved my butt on more than one occasion, although usually I've had software/reformat issues instead of total hardware failure. But I'm due for one of those, I guess.

[JimJ]

Large storage devices (1, 2 or more TBites) have multiple discs as well as multiple readers and suspensions. This all increases the odds of failure. While it is convenient to have large storage devices, the failure risk does increase because of the very reason I mentioned. The more dingle-berries, the higher the odds of failure.

One solution: smaller storage devices (1T or less) or several backups......

If they're all striped as RAID 0, sure. But RAID 1 mirrors the data across two drives, requiring both to fail simultaneously, and RAID 5 would require at least two drives in the array to fail (and the array is still operable with one failed drive).

The Wiki page on RAID has an excellent writeup on this, as well as the logical equations for failure rate with N number of drives.

[jpm]

I had a close call recently with a power outage and a UPS that didn't do what it was supposed to. Two drives on my synology RAID 5 corrupted.  Fortunately one of those was the hot spare.

As a result, I changed to RAID6 (6x4Tb) + 1x4Tb hot spare + 1x4Tb as a separate volume to back up to. I can't back everything up to the internal volume, however I am able to encryot these backups before loading them to Amazon Web Services, with a copy to an external USB drive as well.

All of these drives are Seagates. I run checks monthly and at about 12 months old everything is good.

[Vincent Kars]

I don’t use RAID
It is most of all about high availability, not to be mistaken for a backup.

There are elements in RAID like mirroring that functions as a kind of backup.
However, it protects you against failure of a single HD, if the whole unit got bricked (dropping it, fire, theft, etc) mirroring won’t help you at all.
Striping is a very good technique to improve throughput but if one HD fails, the data is lost.

I prefer having a couple of backups.
I backup to a NAS. This NAS is backed up to a NAS at my sister’s home so I always have an recent copy outside my home.
Half a year I make a copy to an external HD.
As our PC’s don’t have ECC memory, there is a risk of data rot than can go unnoticed for a long time.  Hence to be able to roll back from a backup a half year old might be needed one day.

Most of the time the errors are cockpit errors.
You delete the wrong file, screw up your tags, etc.
I’m not in need of a system storing all these errors redundantly.
I rather have 5 independent disk than 5 in a RAID configuration.

[Wayner]

If they're all striped as RAID 0, sure. But RAID 1 mirrors the data across two drives, requiring both to fail simultaneously, and RAID 5 would require at least two drives in the array to fail (and the array is still operable with one failed drive).

The Wiki page on RAID has an excellent writeup on this, as well as the logical equations for failure rate with N number of drives.

It's the suspension that fails.

[rpf]

Has anyone tried the WD helium filled drives?

http://www.extremetech.com/computing/170213-wd-releases-6tb-ultrastar-he6-the-worlds-first-helium-filled-hard-drive

[k6davis]

Like Vincent Kars, I don't use RAID. It seems like a solution to maximize real-time availability, when what I'm after is a true offline backup (or two) in the event of drive failure. I have no interest in running a redundant copy of my library at all times. That puts the same amount of wear on the "backup" as the main drives get.

Over the years, I have not have good luck with Western Digital. I read the widely publicized article saying that Hitachi drives were far and away the most reliable. Others have disputed the method that study used to arrive at that conclusion, but I've been using them for years and not had a single failure.

[ctviggen]

What brand are you using? I'll stay away from them. I have had good luck so far with Western Digital. I run health checks on them regularly. RAID will eliminate problems of single dsik failure but not of a location based failure such as a house fire. IMHO cloud backup is best for piece of mind.

I absolutely refuse to store any of my data on the cloud.  I have as little other data (such as email) on the cloud as possible. It's incredibly easy for people to steal your information. 

Instead, I have an unraid server and two backups.  One backup I keep out of the house. 

[skunark]

Look's like HGST is the best way to go today.

Looks like the both Seagate and Western Digtial have issues now.

https://www.backblaze.com/blog/hard-drive-reliability-update-september-2014/


Hitachi is now owned by Western Digital so who know what will happen to those drives from a reliability perspective.

More reason to use RAID and/or the cloud for backup.

It you have multiple machines with free space on the hard drives you can use Crash Plan to back them up to each other.

I assume it will take a few years and a new line for HGST before the reliability will be impacted. 

As for RAID, it doesn't necessary mean it's more reliable and for certain configurations it's actually less reliable.  When you do have a single drive failure,  as it's much more likely that a second failure will happen soon.  http://static.googleusercontent.com/media/research.google.com/en/us/archive/disk_failures.pdf  There's other papers that speak to this or tries to argue it, but your best bet is to have multiple backups, along with multiple copies. 

Cloud backups like Backblaze are also not that great of a service, but at least it's cheap.   The problem with the online backup solutions is that they use something called 'inotify' which is an OS feature (windows, mac and linux all have it) to schedule files to be backed up.  This is mostly fine but it's not 100% reliable and leads to another issue as these online services don't offer you a way to rescan your files to do a checksum comparison after they have been backed up. (at least not that i've found)     It it even gets worse,  on a restore from backblaze, they failed to restore several files nor provide any notification of the failed restoration of those files.  One of which I recovered from an old backup and the other two I just re-downloaded the videos from iTunes.  (which i later removed all videos i can redownload from the local drive...)     During this whole process i noticed another music file that had zero bytes so there was some local file corruption going on.   That could have been a telltale sign the drive was going bad, but I wasn't running BTRFS at that time which has a method to check file integrity.   

[skunark]

I absolutely refuse to store any of my data on the cloud.  I have as little other data (such as email) on the cloud as possible. It's incredibly easy for people to steal your information. 

Instead, I have an unraid server and two backups.  One backup I keep out of the house.

It's much more likely to get identify theft from your email vs your music and video being backed up... Which I would argue is impossible   .   For $5/month having one of my computers backup just the media files is rather cheap and a great solution for an automated off-site backup. 

As for key personal files, create a password protected folder, image or thumb drive for safe keeping.     I use a mac which allows you to create a password protected image that you mount like a network drive (.dmg file) which I do allow to be backed to the cloud with an encryption key.   It's much more likely that data will be stollen from the bank, investment or tax service servers than my own, it's probably less likely they will have the data fully encrypted on storage like I do, nor do you want to be the weakest link.   

[kc8apf]

Before I took on a new position at work, I was working directly with all of the major drive vendors on quality, diagnostics, etc. In terms of reliability of modern drives (2T and up), HGST is still the best though they are a bit pricey compared to Seagate who comes in second. There have been a few, specific drive models in the past couple of years that either had design or manufacturing flaws that led to premature failure.

Regarding the # of components in a drive increasing, it's not really the case. The top capacity drives have had 5 platters with 10 heads. The lower capacity versions are the exact same drive with platters and heads removed. So, if you buy a 4T drive today, it has 5 platters while a 2T had 3 or 4 depending on the drive model, yield, etc. What has changed is the components that are most likely to fail. In the past, the mechanical components (motors, servos, actuators) would wear out. The lifetime of the drive could be predicted based on "power on hours" or the # of hours the platters were spinning. Those components have been improved such that they are rarely the first failure. Now, it's the heads themselves that fail, specifically the write head. There is a rough correlation between bytes written and life expectancy. The designs are based on expected workloads for the market the drive is targeting. A desktop drive expects low write rates (<100TB/year) while a server drive expects a lot (>500TB/year). If you buy desktop drives and put them in a NAS, I'd expect early failures.