T
Twayne
In news:[email protected],
Actually, considering the case of running an OS on one, the lifetime
would be seriously cut short of the numbers normally tossed about. And
remember, things like MTBF and write cycles are all statistical
calculations complete with a bell curve and all. Get a component that's
too far to the side of the bell, and it's not going to last long, even
though it carries the same life expectancy numbers.
Anyway: That's because of the constant read/write accesses to the
registry among other system files. Since those writes are going to go
to a consistantly identifiable address set, those locations would get a
lot of writing to them over the course of a day, even if the machine
weren't used. Consider just the clock, for instance, and
I don't know about the 3 years but I do suspect that number isn't too
far off. The "wear" of writing is not an overall number. When you get
it pinned down, EACH address location can be written to so many times.
After that, the location is going to become unreliable and eventually
deteriorate to non-useful, even though overall, the rest of the
locations are going to be just fine, as in, for example, where your own
writings land in the addressing scheme. Those vary and the timing
formulae work.
There are always gotchas to these things.
OTOH, I do think that with some of the technologic advances we're
seeing, that it won't be too long before we see something akin to SSD
being turned into a disk drive with no moving parts. At that point,
alpha migration and all that obscure stuff becomes the life-span
determinants. There are already some types of SSD-like drives available
but it will be quite awhile yet before they see prime time. THAT is
where they become really interesting!
Regards,
Twayne
Bill in Co. typed on Fri, 9 Jan 2009 14:48:01 -0700:
How do you figure Bill? I figured that you have to rewrite the whole
SSD 24 times a day and it would only last 11 years. I am not
including wear leveling which I guess could make it ½ as bad (but
doesn't matter with whole SSD writes anyway). But with me on the
computer all day browsing, email, and newgrouping I only average
about 100MB writes a day. I know this because I buffer all writes to
RAM with MS EWF before commiting. That would take like 22,000 years
to wear out one of my 8GB SSD @ 100,000 complete writes. Or the worst
case with wear leveling acting with every write (which is very
unlikely), 11,000 years.
One manufacture figured out that that the average SSD will last 228
years. Obviously they plan on more writing than I am doing. I do turn
off the pagefile as this greatly decreases my writes and improves
performance. Although I must have enough RAM to make this possible.
There are times that I do write a lot like converting video files
which can be GB worth. Here I throw on an USB hard drive (or network
to one) for these big writes. But I don't do this a lot and it isn't
a big deal to do so.
So while you believe you can wear out a SSD in 3 years, I seriously
doubt that unless you used all of your time to actually try too. Then
and only then, maybe. But most users I don't believe they can wear one
out in 100 years using it normally.
Actually, considering the case of running an OS on one, the lifetime
would be seriously cut short of the numbers normally tossed about. And
remember, things like MTBF and write cycles are all statistical
calculations complete with a bell curve and all. Get a component that's
too far to the side of the bell, and it's not going to last long, even
though it carries the same life expectancy numbers.
Anyway: That's because of the constant read/write accesses to the
registry among other system files. Since those writes are going to go
to a consistantly identifiable address set, those locations would get a
lot of writing to them over the course of a day, even if the machine
weren't used. Consider just the clock, for instance, and
I don't know about the 3 years but I do suspect that number isn't too
far off. The "wear" of writing is not an overall number. When you get
it pinned down, EACH address location can be written to so many times.
After that, the location is going to become unreliable and eventually
deteriorate to non-useful, even though overall, the rest of the
locations are going to be just fine, as in, for example, where your own
writings land in the addressing scheme. Those vary and the timing
formulae work.
There are always gotchas to these things.
OTOH, I do think that with some of the technologic advances we're
seeing, that it won't be too long before we see something akin to SSD
being turned into a disk drive with no moving parts. At that point,
alpha migration and all that obscure stuff becomes the life-span
determinants. There are already some types of SSD-like drives available
but it will be quite awhile yet before they see prime time. THAT is
where they become really interesting!
Regards,
Twayne