Re: Life Expectancy of Flash Memory Drives
From: Stephen Moll (stephen.moll_at_nospam.amsjv.com)
Date: 07/01/04
- Next message: angusyoung: "compiling external modules"
- Previous message: David Kinsell: "Re: DDS4 Tape Errors"
- Next in thread: Kenneth Lantrip: "Re: Life Expectancy of Flash Memory Drives"
- Reply: Kenneth Lantrip: "Re: Life Expectancy of Flash Memory Drives"
- Reply: Daniel Ganek: "Re: Life Expectancy of Flash Memory Drives"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Thu, 1 Jul 2004 15:20:02 +0100
"Kenneth Lantrip" <nunya@bidny.ok> wrote in message
news:0nAEc.15407$J82.10525@fe25.usenetserver.com...
> Yesterday, my PNY Atache flash memory stick drive started to fail on me.
> It can read all files just fine. So I managed to tar all the files into
> a backup.
>
> I can reformat it in Win2000 or Knoppix just fine. But when putting
> files (of any kind) onto it will lock up the computer (W2k or Knoppix)
> every time after about a minute or so of writing to the device. It does
> this on two different computers that I've tried.
>
> Conclusion: The drive is bad.
>
> My query is: Does anyone know about how long one should expect one of
> these type drives to last under daily use as a Knoppix persistant home
> directory? Anyone know of any utilities to lock out a bad memory
> cell/sector?
>
> I thought since these things don't have moving parts (other than moving
> from place to place) they should last quite a while and be very reliable
> from data loss. Guess I was wrong.
>
> Anyone else run into problems with flash memory?
>
> Thanks.
>
> Kenneth Lantrip
>
This type of device is almost undoubtedly going to be NAND based flash
memory. This type of memory is broken into blocks of 8 to 32k, each of which
has to be erased before it can have data written to it. Each block has a
life expectancy of 100,000 to 1,000,000 erase cycles. On top of that there
is a general MTBF (Mean time before failure) of around 500,000 hours.
What this means is that after 100,000 erase cycles for a particular block,
there will be an ever increasing likelihood that block will start creating
errors. By 1,000,000 it will almost certainly have stopped working. Even if
none of the blocks in your memory device have had less than 100,000 erase
cycles, the device as a hole is expected to fail at some point, with the
highest probable failure time in 500,000 hours (over 50 years).
Remember that these figures are based on probabilities, and a manufacturer
can only really say that ninety-something percent of devices will meet these
figures. There will be a small percentage that fall outside either by
failing earlier than expected, or if you're lucky... later.
All of these devices have defects in them when made, and part of the
manufacturing process involves something akin to formatting where bad blocks
are marked as bad on the chip. Some may pass the test at manufacture only to
fail soon after. For mulitmedia storage this isn't usually a problem as
small numbers of bit glitches are lost in the noise, but for program storage
this can be a problem... obviously.
So it really depends on what you use it for. I have known people use them as
system storage on PDAs, running alernative operating systems from them, only
to be surprised when the system becomes unstable after a few weeks. I have a
USB stick, which I've had for about a year. I actually store software on it,
but the contents are rarely changed, only once or twice a month, and it is
still fine.
Do you use it heavily?
- Next message: angusyoung: "compiling external modules"
- Previous message: David Kinsell: "Re: DDS4 Tape Errors"
- Next in thread: Kenneth Lantrip: "Re: Life Expectancy of Flash Memory Drives"
- Reply: Kenneth Lantrip: "Re: Life Expectancy of Flash Memory Drives"
- Reply: Daniel Ganek: "Re: Life Expectancy of Flash Memory Drives"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|