I had a long talk with an engineer friend who works in flash memory. It wa=
s
geeky but illuminating.
Each flash memory cell is a small capacitor. One interesting fact is that
in order to read it, the bit must be actively flipped. And reading disrupt=
s
other cells in the same column. In fact, in the process of reading, the
memory is re-written. Cells have a lifespan and begin to degrade after a
certain number of cycles. There is a considerably large amount of error
going on almost all the time, but it's a generally predictable degree and
type of error, so data is written with a degree of redundancy using error
correcting read/write algorithms. A microcontroller keeps track of which
cells or areas are experiencing higher degrees of error and as they begin t=
o
fail they are permanently blocked out.
Personally, I trust flash memory long-term about as much as any other
medium, which is to say I have reservations. I believe the only best
practice involves redundant copies stored in disparate locations, and
periodic verification and re-copying.
-jeremiah
On Sat, Jun 26, 2010 at 3:57 PM, Mitch Hill <> wrote:
>
>
> At 10:35 PM 6/26/2010 +0000, you wrote:
> > it is
> > > still possible for any given specimen to fail within the first hour
> > > of use. It's just statistically less likely.
> >
> >
> >Like a light bulb, most failures seen to happen during a transition such
> >as from unpowered to powered.
> >
> >I seem to remember some years ago a company was developing a way to
> >preserve data onto paper. I believe they were able to transcribe the 1's=
&
>
> >0's to print on archival paper. And a way to accurately scan that back
> >into digital.
>
> Actually, keeping data on paper for archival purposes has been around for=
a
>
> long time. And I'm not talking about the bible...
>
> Way back, in 1958-59 that I know of first hand, my father in his business
> used punched digital paper tapes and punch cards for data storage and
> retrieval. A duplicate set was maintained in a bank vault in town and
> weekly, updates were made and swapped with the ones in the vault. I
> believe this method dates back into the late 1940's, was very labor
> intensive and costly and we've come a long ways since then...
>
> But I can still hear the chatter of the mechanical Frieden (sp?) tape pun=
ch
>
> chugging along to this day...
>
> >------------------------------------
>
> >
> >"While a picture is worth a thousand words, a
> >sound is worth a thousand pictures." R. Murray Schafer via Bernie Krause
> >Yahoo! Groups Links
> >
> >
> >
>
> --
> TNX, 73, Mitch Hill - K1FH
>
>
> http://www.4shared.com/dir/UTASxktL/wildlife.html
>
> Shadow's area: http://www.4shared.com/dir/ecfWjyZb/Shadow.html
>
>
>
>
--
-----------------------------------------------------------
jeremiah moore | SOUND |
http://www.jeremiahmoore.com/
|