A friendly warning ...

[Bjørn Rørslett]

I beg to disagree. A backup strategy is less complex than the digital camera we take for granted, and their manuals. Basically it boils down to having more than one copy of any valuable data and knowing where to retrieve and restore it. Whether this aim is achieved by a script, canned OS procedure(s), or a third-party software package, is immaterial. A lot of options exist and cost from nothing to much less than a camera.

Verifying the backup tape(s) is an automatic procedure that needs virtually no user intervention. You can use disk-disk backup (virtual tape) in the same manner if tape is not feasible.

[Jan Anne]

I think I have around 1TB of images which isn't very much so I'm thinking of a single NAS with enough disks that one or maybe two can crash without any dataloss.

As flooding will be a major concern in my new place I intend to have a couple of off site copies lying around which will be exchanged for a fresh copy whenever I visit said location. In the future when faster Internet is available on my remote location I might co invest with others to have two or three NAS boxes synced online. Using the cloud might be another option but I'm not a belieber in the concept that others can determine what they can do with my data.

[Andrea B.]

In perhaps 25 years of using home PCs and laptops I have had one (1) disk crash.
Bjørn, I cannot understand how you accumulate so many! Maybe you need some new machines to run them? Something is going on there.

[Bjørn Rørslett]

Crashes tend to follow a clustered extreme value distribution. So there can be many months without any, then all of a sudden several in a row. Some disks die very young, whilst most give up the ghost after running for years.

The number has to be seen relative to the population size,  thus failure rate is not much different from other mass production items, photographic lenses included. Actually, single-point estimates of lens failures are higher.

[PeterN]

I switched to cloud service. No worries about backup, hardware crashes......I hope

[Jørgen Ramskov]

I beg to disagree. A backup strategy is less complex than the digital camera we take for granted, and their manuals. Basically it boils down to having more than one copy of any valuable data and knowing where to retrieve and restore it. Whether this aim is achieved by a script, canned OS procedure(s), or a third-party software package, is immaterial. A lot of options exist and cost from nothing to much less than a camera.

Verifying the backup tape(s) is an automatic procedure that needs virtually no user intervention. You can use disk-disk backup (virtual tape) in the same manner if tape is not feasible.

I think you misunderstand me. A backup strategy isn't complex for sure, but making sure you have a backup of everything is as it is often spread out to many places. My primary point is probably that it's something people simply don't care about until it's too late. I think it's a problem that a company like Apple could really take to the next level.

[Jørgen Ramskov]

I think I have around 1TB of images which isn't very much so I'm thinking of a single NAS with enough disks that one or maybe two can crash without any dataloss.

As flooding will be a major concern in my new place I intend to have a couple of off site copies lying around which will be exchanged for a fresh copy whenever I visit said location. In the future when faster Internet is available on my remote location I might co invest with others to have two or three NAS boxes synced online. Using the cloud might be another option but I'm not a belieber in the concept that others can determine what they can do with my data.

The remote backup I have in place is exactly like that. My brother and I am fortunate that we both have fast, fibernet based internet connections, so we're simply running remote backups to eachothers NAS.

I'm curious why you'd choose to move to a place where flooding will be a major concern?

[Jørgen Ramskov]

In perhaps 25 years of using home PCs and laptops I have had one (1) disk crash.
Bjørn, I cannot understand how you accumulate so many! Maybe you need some new machines to run them? Something is going on there.

He's got a lot of harddrives running, he's bound to experience a lot more crashes. Having worked in IT outsourcing, I have witnessed how often even server disks and hardware fails.
Another thing I read several years ago is that I believe it was Google, had done some RAID studies on a massive scale and warned about using RAID5. The reason was the size of the disks made a resync of the RAID take quite a lot of time and their statistics showed that there was too high a risk that a second disk would fail before the failed disk had been replaced and the RAID back in sync. Google don't use RAID at all I believe, they have created their own filesystem and I believe it automatically makes sure that all data is located in 3 different places at all times (a very simplistic view, it's of course much more advanced than that). 

[Jørgen Ramskov]

I switched to cloud service. No worries about backup, hardware crashes......I hope

Interesting, could you share some more details about your setup?

[Bjørn Rørslett]

It is correct that the bit rate error of consumer disks with their current huge size makes building vast RAID5 volumes more risk-prone than earlier. Hence it is wise to take the array off-line during the rebuild and have additional spare(s) at hand. I have yet to encounter a failed rebuild due to bit errors, but this is sooner or later going to happen for sure.

There are alternate RAID configurations which may suit today's massive space requirements better than RAID5 in terms of security, but not when economy is factored into the equation as such RAIDs generate more overhead for additional parity and striping disks. Up to around 10 TB should still be feasible for consumer-class RAID5 systems though.

[Jørgen Ramskov]

RAID6 would be the obvious next step. It's depends on now critical the system is and the discussion isn't relevant for many of the most critical systems - they are connected to a SAN instead.

We live with the possibility of hardware failure in our cameras too - flash cards fails regularly too, which is probably why it's possible to create a mirror with two cards in many cameras.

[Bruno Schroder]

Data Storage on DNA Can Keep It Safe for Centuries:http://www.nytimes.com/2015/12/04/science/data-storage-on-dna-can-keep-it-safe-for-centuries.html?_r=0

Not yet available unfortunately

[frankv]

The fact that you had your brother (who works in IT) to configure or making sure your had a proper backup solution just proves my point: It's way too difficult.

It's because of him that I decided to work out a backup strategy. But I set it up myself. It's not terribly advanced, but it suits my needs. I have learned that a verified backup is important, after a harddrive failed and I had to recover a day from my (not verified) backup. No need to tell you all it failed. :-)
So a revision of my strategy was needed.

I use a free tool called Cobian, and cycle two external harddrives, one running backup, and one stored outside the house. I swap them around every two weeks or so.


-frank-

[Michael Erlewine]

If you have a Blu-Ray DVD writer, it will write to M-Disks (Millennial disks), which are said to last 1,000 years if properly handled. A Blu-Ray M-disk will store 25 GB of data, which is not too bad. More info here:

http://www.mdisc.com/

[Bjørn Rørslett]

Unfortunately, although the media itself just might survive a millenium, the chance that the reader will do the same is virtually nil.

Thus we push the problems ahead of us.