On 2010-06-07, DoN. Nichols wrote:
On 2010-06-06, Michael A. Terrell wrote:

Wes wrote:

"Michael A. Terrell" wrote:

There is no moving mechanical mass in a solid state drive so it's a
lot faster that conventional drives. there is no oxide to shed from the
spinning media, and they don't mind vibration or suffer from mechanical
shock like older drives.

But the flash cells have a finite life as in write cycles.


It was well over 100,000 write cycles the last time i worked with
solid state drives. Since the drive in a machine tool is mostly a read
device, it should outlast the tool. NASA had no problem with the
expected life span in 2000.

It depends on the OS. Unix variants (unless the filesystem is
mounted read-only) update the last-accessed time in the inode for each
file every time you read it. And there are a number of file inodes
sharing the same 512 byte block, so the number of write cycles in an
active directory would accumulate rather faster than you might expect.

In Linux, there is a mount option "noatime" that disables that
behavior.

This could at least be a problem with EMC (which is linux based)
-- unless care is taken to mount the SS drive read-only, and keep some
other form of medium for read-write. Perhaps set it up using the SS
disk something like a liveCD, and copying it into a RAM-DISK for normal
execution, and only remounting it read-write when you absolutely need to
write something to it.

With the Microsoft FAT filesystem, I don't think that there is
any equivalent last-accessed time, but I don't know what other gotchas
might be present.

This is not a big problem, one is because it does not result in that
much writes, and also because of availabiilty of noatime.

i