Using CF cards like film

[Wayne Fox]

I know this has been discussed before, but economics have changed again.  I just bought 16 4gig cards (not the fastest but fast enough for what I shoot) at costco for an upcoming trip for $12.50 per card.  I'll be backing the cards up daily, but planning on not reusing a card during the trip. I know others are doing this with the low price of cards now.

However, at this price it means the card as long term storage is now much less than I used to pay for film and processing.  I'm considering just writing the date and other info on the back of the card and never reusing one at all ... just locking them away as a last resort backup.  I've been stung with cards failing after a time so this problem wouldn't occur, but not sure if these low price cards (still made by SansDisk) are perhaps susceptible to problems.

I know hard drives are still cheaper, but this is pretty cheap.  Maintaining hard drives and hard drive backups is pretty reliable, but I know if you lock one up for a few years or more, it can have problems if you try and mount it again.  

Perhaps CF cards aren't any better ... and I can't find any info.  It was discussed last year, but no conclusions as to how reliable is a CF card that has been shot only once and then stored?

[Gemmtech]

I just don't believe there's enough reliable data to make an educated statement.  I remember when CD-Rs first hit the market with claims of 100-200 years everybody felt secure, then the unthinkable happened and they started to disintegrate.  I had great luck with Kodak Golds and to this day never had a failure, the same for Mitsui (now called Mam-a).  Needless to say I back up ALL my data and I believe CF needs to be treated the same.  AMAZING, negatives seem to last forever and now we need to backup and backup and then backup again not knowing if our computers will be able to read the files in 20 years, but we know the negatives will still be good.
I think it's a wonderful idea, shoot a CF and save it, makes it very easy.  SS type devices tend to have problems after they have been rewritten to many times, shooting once and saving is probably the best guarantee that everything will be OK.

[PierreVandevenne]

Perhaps CF cards aren't any better ... and I can't find any info.  It was discussed last year, but no conclusions as to how reliable is a CF card that has been shot only once and then stored?

I started writing a longish response to that, but then I realized it was full of digressions. So to cut a long story short... I have been involved in data recovery, one way or the other, for the last twenty years and that's exactly what I do when I am on a trip. Note: while I have yet to see an inactive CF card suddenly fail after prolonged storage, I have quite a few horror stories about airports, which in theory shouldn't happen, but do happen. Long term (as in more than ten years) is another issue though as we don't have real data.

Major brand cards are now rock solid, provided you aren't at the cutting edge (4GB-8GB). I'd be more careful with 16 and 32GB UDMA cards and new cameras at this point.
You'll of course still need to download them to your computer and use an external hard drive as a backup of the computer as an insurance against other issues (lost luggage, fall in water, etc...)

There's one caveat though: initial card failures are uncommon, but not outlandishly rare. I use the cards to full capacity at least once with the cameras I am taking on the trip  (normal shots and a few burst mode runs) before leaving for the field..

[Jonathan Wienke]

I'd say that doing due diligence to test the card is wise, but once the card has a track record of storing data, there's no reason to believe flash memory isn't at least as reliable as a HDD for long-term storage. Other than their connector pins, flash cards have no moving parts to wear out, no bearings to seize up, no read/write heads to crash into a spinning platter, etc. A flash card is far more likely than a HDD to survive submersion, prolonged exposure to high humidity, extreme temperatures, or high G forces. The flash cells are subject to a finite number of write cycles before failure, but they are sealed inside the chip packaging and there's no limit to the number of times data may be read from the cells. Flash cards also generate far less heat than a HDD.

Most unrecoverable HDD data loss is caused either by a mechanical problem (head crash, bearing failure, condensation on drive platter, etc.) or component failure due to excessive heat buildup--things that don't really apply to flash memory devices. The only thing that might keep flash from trouncing HDD reliability-wise as an archival medium is some as-yet-undiscovered factor causing chip failure after several decades of storage. But I haven't seen any evidence so far to indicate such a factor exists, so overall I'd say it's safe to assume you're no more likely to lose data archived on flash cards than on hard drives.

[Wayne Fox]

I'd say that doing due diligence to test the card is wise, but once the card has a track record of storing data, there's no reason to believe flash memory isn't at least as reliable as a HDD for long-term storage. Other than their connector pins, flash cards have no moving parts to wear out, no bearings to seize up, no read/write heads to crash into a spinning platter, etc. A flash card is far more likely than a HDD to survive submersion, prolonged exposure to high humidity, extreme temperatures, or high G forces. The flash cells are subject to a finite number of write cycles before failure, but they are sealed inside the chip packaging and there's no limit to the number of times data may be read from the cells. Flash cards also generate far less heat than a HDD.

Most unrecoverable HDD data loss is caused either by a mechanical problem (head crash, bearing failure, condensation on drive platter, etc.) or component failure due to excessive heat buildup--things that don't really apply to flash memory devices. The only thing that might keep flash from trouncing HDD reliability-wise as an archival medium is some as-yet-undiscovered factor causing chip failure after several decades of storage. But I haven't seen any evidence so far to indicate such a factor exists, so overall I'd say it's safe to assume you're no more likely to lose data archived on flash cards than on hard drives.

Thanks.  Sort of sums up the logic I've gone through in thinking about this.

I'm not making any plans to change my current HDD strategy including backups.  More like a new backup step, since I'll keep the cards locked in a separate location ... as I mentioned a backup of last resort type of thing.

[DarkPenguin]

Thanks.  Sort of sums up the logic I've gone through in thinking about this.

I'm not making any plans to change my current HDD strategy including backups.  More like a new backup step, since I'll keep the cards locked in a separate location ... as I mentioned a backup of last resort type of thing.

Let us know how long they last.  My big concern would be that no error recovery is being done if the cards are not being accessed.  (Hard disks will move data if a sector is becoming unreliable.  That doesn't happen if the sector is never read.  I'm not sure what mechanisms exist for flash cards to do the same.)

You could find something that will checksum the files on the cards.  Toss that checksum on the card.  Then, randomly, take cards out and check the checksums.

[PierreVandevenne]

Let us know how long they last.  My big concern would be that no error recovery is being done if the cards are not being accessed.  (Hard disks will move data if a sector is becoming unreliable.  That doesn't happen if the sector is never read.  I'm not sure what mechanisms exist for flash cards to do the same.)

You could find something that will checksum the files on the cards.  Toss that checksum on the card.  Then, randomly, take cards out and check the checksums.

I've just run checks on dozens of cards I have in storage since 2001 and have found no errors. This being said, I dug a bit deeper by reading datasheets and a few research papers on the topic. The floating gate will definitely leak over time, but it is difficult to give exact figures for many reasons (lack of standard definitions and testing procedures, difference in types of cells (SLC/MLC), overoptimistic figures by the NAND producers, overpessimistic figures by the NOR producers a few years ago and the conventional hard drive industry today, differences in usage, difference in ECC algorithms, storage temperature, the system as a whole with ECC and bad block management, etc...). The typical figure given by the NAND flash industry is around 10 years at 10% P/E cycles for the chips themselves and 50 to 80 years for the systems themselves, but the trend seems to be moving towards shorter rentention time. I haven't seen any specific figures for data retention for totally unused NAND flash (you can always extrapolate usage/data retention curves to the extreme, but how meaningful is tthat). Some use seems to be good because it allows the system error correction mechanism of the whole system to intervene before too many bits are lost. Too much use is bad because it worsens leaks (and other cause of errors). Last but not least, heat never helps, as usual with electrons and silicon.  Anyway, it is quite clear that it won't last "forever" because of leaks. A somewhat balanced summary here.

http://www.slideshare.net/siliconarrow/Rel...tate-Drives2008

PS: cheaper cards are probably MLC

[Deep]

For what it's worth, my oldest compact flash card (32MB, bought in late 2002) failed a few months ago, with very few read/write cycles.  I had used it as backup storage on an HP Jornada mini-laptop but later reformatted it for a camera.  Now it can't be recognised by any camera or card reader I have and has been relegated to museum piece.  So they CAN fail, like anything else.

Don.