Is there a foolproof way to store your memories?

[solofly]

http://technology.canoe.ca/2010/02/24/13009136-cp.html

Good article...

 

[Blain]

Most of the time people don't pay any attention to the quality of blank media when they make a purchase.
I would estimate that AT forum members give a little more thought to "quality" when they buy, but not much.
Many people burn a CD or DVD, if it works good for the next week, they proclaim it "quality media".

 

[C1]

For me HDDs are still the preferred media, but I have read that DvDRAM is also very good. There is also special archive grade/quality DvD available for backup & I would recommend at least that grade or type media for any serious long term storage (eg, 5-10 years).

But remember in all of this, stuff doesnt stay compatible long. Eight inch floppies=>51/4 floppies=>3.5 floppies=>CD=>DvD=>BlueRay=> and now there is report of being able to put 500Gb on a BlueRay type/size media. So after 10 years, you will be required to rearchive your media or you will lose the ability to read/use it anyway (& that includes HDDs). We see this problem already even once in awhile in these forums for people needing 51/4 floppy readers.

 

[Fayd]

HDD's with offsite backup seems to be the best option. as for what form that offsite backup takes, let's say you have a safe deposit box at the bank, unless your neighbors are storing magnets, you're probably fairly safe storing an external harddrive in foam packaging.

 

[Modelworks]

Usb memory stick kept inside a metal box will last a very very long time. And if that isn't good enough and it is really important data then you can get PROM devices which physically burn out fuses inside the chip to record the data. Life expectancy is around 2000 years.

 

[VirtualLarry]

Usb memory stick kept inside a metal box will last a very very long time. And if that isn't good enough and it is really important data then you can get PROM devices which physically burn out fuses inside the chip to record the data. Life expectancy is around 2000 years.

Not really. Lifetime of MLC flash is what, 5 years? At least that's what the specs for SSDs say. The new 34nm, 25nm, and 3LC flash are shorter than that. I wouldn't trust flash for long-term storage. Quite frankly, I would trust a HD for longer than flash memory.

 

[Modelworks]

Not really. Lifetime of MLC flash is what, 5 years? At least that's what the specs for SSDs say. The new 34nm, 25nm, and 3LC flash are shorter than that. I wouldn't trust flash for long-term storage. Quite frankly, I would trust a HD for longer than flash memory.

memory for drives is rated at so many reads/writes over a 5 year period. That isn't how long it retains the data if you write to it and store it away somewhere. Once you write the data then store the drive the only things that will cause it to lose data are time ( the state of the cell will change over long periods) & external factors (temperature, gamma rays, magnetic fields) . The current thinking is that a new flash USB drive used only once to write the data to it and sealed in a metal enclosure, kept cool and dry, could last upwards of 20 years or more. Nobody can tell just how long because it hasn't been around long enough.


Hard drives are a very bad choice for long term storage. The interface and file system is unique to the time when the data was written to the drive. I can't imagine trying to find a sata controller 10 years from now and an OS that can read the drive. Try to find an MFM or RLL controller and a pc to use it with the correct drivers and OS and you will see what I mean.
Serial and parallel flash can be accessed by any digital computer, nothing special required.


PROM is the way to go if you really must have something preserved in digital form . It never loses its contents. If you don't need it in digital form then the best way to preserve information is old tech, paper

 

[jimhsu]

10 years is the minimum "acceptable" period for data retention for consumer grade flash. However, most issues of data retention arise from issues of interfaces and the encodings used, not the physical media itself. Try extracting useful data from a 1950s tape drive, even with a completely working machine. Do you really think that ZIP file will be readable 30 years from now? Or that you can buy a Blu-ray player at that time?

"Without a migration plan, I would say that a lot of (digital home movies and photos) will be lost, but not necessarily only due to deterioration of the storage media," Slattery writes. "While the bits and bytes may still be intact on a particular medium in 30 years, accessing that data may be difficult."

No, not paper. Most^H^H some paper is not acid free and storage beyond a couple of decades is ... unlikely due to other factors such as the ink used.

Stone/metal engraving FTW. That can literally last millenia if undisturbed. There's a reason why most "alien artifacts" seen in movies are engraved or etched on a solid surface. Also the presence of data is unambiguous - there is either something scrawled on the surface, or not. Magnetic and optical media -- not so much.

There is no digital alternative, other than constant data rotation to newer formats. That's probably the only way data can be stored indefinitely.

 

[jimhsu]

Just found this in the first anandtech article about the Intel X25-M:

It's also possible for a flash cell to lose its charge over time (albeit a very long time). Intel adheres to the JEDEC spec on how long your data is supposed to last on its SSDs. The spec states that if you've only used 10% of the lifespan of your device (cycles or GB written), then your data needs to remain intact for 10 years. If you've used 100% of available cycles, then your data needs to remain intact for 1 year. Intel certifies its drives in accordance with the JEDEC specs from 0 - 70C; at optimal temperatures your data will last even longer (these SSDs should operate at below 40C in normal conditions).

 

[VirtualLarry]

Once you write the data then store the drive the only things that will cause it to lose data are time ( the state of the cell will change over long periods)

Yes, and that period of decay is around five years, for current-gen flash chips. Next-gen flash chips will have even shorter longevity. Factors include the physical size of the cell, and how many bits stored per cell. Newer flash memory with smaller physical cells will have more electron leakage, and lifetimes are going to be much shorter. Future controllers will have to deal with limited lifetimes by intentionally re-writing static stored data on drives, otherwise errors will occur. This cannot happen if the drive is powered off and stored for multiple years.

Hard drives are a very bad choice for long term storage. The interface and file system is unique to the time when the data was written to the drive. I can't imagine trying to find a sata controller 10 years from now and an OS that can read the drive. Try to find an MFM or RLL controller and a pc to use it with the correct drivers and OS and you will see what I mean.
Serial and parallel flash can be accessed by any digital computer, nothing special required.

I disagree. HDs use a standardized interface. So do USB flash drives. Either one is subject to becoming obsolete, there's nothing magical about flash drives in terms of interface longevity.

When you say serial and parallel flash, I hope you're not referring to having to remove the raw flash chip die from the USB flash drive and reading it out directly, are you? Because then you have to have intimate knowledge of the controller that was paired with the flash chip, in order to read out the high-level data. Far, far more complex than just keeping something with a USB port around.

It's highly unlikely that one would not be able to obtain an adaptor for an older interface standard.

MFM and RLL were not controller interface standards, they were raw data-encoding standards. Those drives were more or less permanently attached to their controllers, you can't even pull a drive off of one controller and read it on another. Much like pulling a flash chip out of a flash drive and attempting to read it from another controller chip.

Whereas, any IDE, SATA, or USB drive can be unplugged from one machine and read easily on another, because the controller is bound to the drive, and has a standardized system interface.

Edit: Read this:
http://www.patriotmem.com/products/...p=Torqx%20M28 Solid State Drives&catid=21
"Data retention: 5 years at 25C".

Considering most case temps are elevated, it stands to reason that 5 years for your data is on the longer side of the estimates.

 

[Fayd]

oh.... one thing to mention.

one company i worked at specialized in off-site data storage. technically, their job was protection of intellectual property... contracted records retention, i guess. anyways, pretty much everything they had was stored on 5.25" magneto-optical disks, 5.2 gig capacity each. HUGE bags of these disks, each in their sleeves, with a number to mark them. several bags into a box... several boxes onto a rack... hundreds of racks to a vault.

 

[nyker96]

maybe carve it into stone? if it's good enough for the ten commendments, I think it should be good enough for my memories :]

 

[Modelworks]

Yes, and that period of decay is around five years, for current-gen flash chips.

I don't know where you get the idea that flash chips are 5 year parts. Maybe the ones used in SSD have that limitation, but flash used in other devices and nand devices has at least 10 years and sometimes 20 years guarantee from the manufacturer. Nanya has many products that are in those ranges as does Atmel. I have flash parts from 1989 that can still be read.

I disagree. HDs use a standardized interface. So do USB flash drives. Either one is subject to becoming obsolete, there's nothing magical about flash drives in terms of interface longevity.

HD interface is common to other HD. It isn't something you can use without having extensive prior knowledge. USB flash drives have the advantage the interface is simple to reverse engineer if needed.

When you say serial and parallel flash, I hope you're not referring to having to remove the raw flash chip die from the USB flash drive and reading it out directly, are you? Because then you have to have intimate knowledge of the controller that was paired with the flash chip, in order to read out the high-level data. Far, far more complex than just keeping something with a USB port around.

The controller chips used in usb drives use a very basic serial protocol. Even someone starting from no knowledge about the specific controller used but with knowledge the protocol is serial can retrieve the data. I don't need to know the file system used or what OS put it there all I have to do is ask the controller for the data . I don't have to issue specific commands to control the drive. I can do a raw dump of the data at the very lowest level of hardware access with just a few I2c or SPI commands which are universal to all devices that use them, not just storage.

MFM and RLL were not controller interface standards, they were raw data-encoding standards. Those drives were more or less permanently attached to their controllers, you can't even pull a drive off of one controller and read it on another. Much like pulling a flash chip out of a flash drive and attempting to read it from another controller chip.

The point is recovering data from them would be difficult now because the technology is dead. Try to pull data off a 8" floppy now . IDE and SATA will be the same way in the future . Even if you could interface to the drives you will have to know how the protocols work and the proper command addresses to issue. Sending the wrong commands can erase data or corrupt the drives firmware. With flash controllers I can pull the data off bit for bit.

Whereas, any IDE, SATA, or USB drive can be unplugged from one machine and read easily on another, because the controller is bound to the drive, and has a standardized system interface.

That means nothing if you have no pc with IDE, SATA , or USB in the future. Then what matters is having a device with the simplest interface to the data as possible.

Considering most case temps are elevated, it stands to reason that 5 years for your data is on the longer side of the estimates.

There are flash memory out there that has been used for longer than 5 years in much more hostile conditions than inside a pc case. I think you are confusing SSD drives with flash memory in general.

 

[CurseTheSky]

The best way to store data is to:

1. Keep it in multiple places at the same time.
2. Upgrade the storage media and interface as time goes on.

It's painstaking and expensive, but if your data is truly worth that much to you, it's an acceptable burden. Relying on a single format to last decades is ludicrous. Yes, we may have CDs, DVDs, Blu-rays, tape drives, HDDs, flash memory, etc. that can still be read 50+ years from now, but is there much of a point? By that time, the data should either be invalid / not needed, or it'll be extremely difficult or expensive to find a machine capable of retrieving the contents. Not to mention newer technology is growing in reliability, speed, and capacity every day.