Flash Hard Drives

[Charles Steinkuehler]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Jon Pruente wrote:
| On Jan 13, 2008 4:58 PM, Earle Beason <Earle-Beason at kc.rr.com> wrote:
|>     There are several companies that make solid state hard drives, I was
|> able to play around the samples, they seem to respond quicker then the
|> disc hard drives.
|>
|>      I had a concern on the sustainability of memory over the lifespan
|> of the device, most flash chips I am aware of have a read/write life
|> span of 100,000 cycles. I am just wondering how long will it actually
|> last before their is a failure
|
| Most flash controllers do have a reserve of blocks for bad block
| management so even if a block suffers a failure you might not notice
| as it will be remapped.  I would think that though an individual block
| might have a life span of 100,000 writes (which is the low end of
| current hardware as I understand) it will be a very long time until
| the device runs out of reserve blocks.

Particularly since pretty much any flash based device like a HDD, CF
card, SD card, etc. does hardware based wear-leveling these days.  The
overall lifetime is a function of the device size, the minimum
guaranteed number of rewrites (and 100K is towards the low end these
days), and the rate at which you're writing to the media.

If you're using raw flash chips in an embedded product you do have to
worry about sector wear, but that's what all those flash file-systems in
the kernel are for.  :)

- --
Charles Steinkuehler
charles at steinkuehler.net
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHirPpLywbqEHdNFwRApvwAKC98oHLWs5661/VC4qnH6BqJdX4+gCeOLeF
Aw7Mzs2ulqbFg1amMNFumSw=
=NFsn
-----END PGP SIGNATURE-----