If you're worried about a HD failing to the point it destroys your data,
One bit dropped will ruin a photo file. HDs -- as they come off the line and are certified for shipping -- fail to the point of destroying data. They ALL do it, early and often. The drive circuitry has error correction built in that usually corrects the trouble, but there is a threshold, and once that threshold is passed, your photo is destroyed for all time.
The tighter the data-packing, the greater the need for error correction and the more imminent the threshold limit becomes.
[!--quoteo(post=246850:date=:name=from giles)--][div class=\'quotetop\']QUOTE (from giles) [a href=\"index.php?act=findpost&pid=246850\"][{POST_SNAPBACK}][/a][div class=\'quotemain\'][!--quotec--]1. keep copies of your data on multiple drives
2. keep at least one copy of everything off site
3. update the disks from time to time, especially as connectivity options change (SCSI, IDE -> SATA, firewire seemingly on its way out, USB3 is on the horizon, eSATA may yet find some market share, who knows what next)
4. yes, #3 implies that even old "archived" images need to be migrated to new media
5. checksums to be sure that the data you stored is the data you read back are a good idea (sometimes disks lie), and painful to manage manually so I hope that Apple and Microsoft enhance their file systems along the lines of what Sun has done with ZFS[/quote]
Did I mention that HDs are a PITA to manage?
I, too, have never had an HD fail. I have, however, had a lot of files corrupted one way or another in ordinary operation. Something as simple as moving a file from one directory to another (which, back in the old days, involved only the changing of a pointer in the FAT) has caused problems. Disk optimization can bring on corruption. Simple age will cause the magnetic bounds to spread and the fields to weaken, increasing the error rate. Disk expansion/contraction from heat introduces tracking errors, exacerbated by the incredibly narrow tracks in use today. And so on.
In line with #5 above, I surely wish that MS (IBM is really to blame, but MS is easier to pick on) had included at least checksum validation in its file copy/transfer protocol. But they didn't, haven't and most likely won't, so we must rely on 3rd party ware to do that for us, and that too is a PITA.
As for CD/DVD storage: The medium of an RW disk is a kind of jelly sandwiched within plastics and garnished with a thin (nearly transparent) aluminum sheet. The jelly is only semi-rigid, and is compacted when heated by the writing laser, leaving little (I mean
tiny) pits and bumps, each one a bit, along a track. When read, the laser heat is less, so it supposedly has no effect on the jelly. Even so, over time (leaving aside oxidation of the metal), the jelly spreads, as any goo will, and begins to refill the pits. After awhile: data destruction. The only real question is the length of the while. Gold foil (because it doesn't oxidize) increases longevity. Certain jellies are said to resist spreading longer than others. But... how long before CDs, et al., go the way of ten inch floppies (or half-inch tape) and become unreadable by way of technological change, regardless of data longevity?
My question, in the beginning (rephrased, but unchanged in intent): Has anyone tested long-term data integrity on flash media? So far, it looks like NOT is the answer.