Solid State Drive Fragmantation?

[cameo]

Can solid state drives also slow down over time like traditional hard
drives due to fragmantation or some other reason that is specific to
solid state drives?

 

[Scott]

Can solid state drives also slow down over time like traditional hard
drives due to fragmantation or some other reason that is specific to
solid state drives?

Those more expert than me can provide a definitive answer but my
understanding is that fragmentation is not an issue with SDD and
indeed you should turn off defragmentation (and indexing) as such
processes may shorten the product life.

 

[Andy Burns]

Can solid state drives also slow down over time like traditional hard
drives due to fragmantation or some other reason that is specific to
solid state drives?

They can, but don't run a traditional de-fragmenter on them, that can
shorten their life and make things worse.

Should we presume you're running Win7? If it recognises the drive as an
SSD, it should enable TRIM support and disable the inbuild defrag
scheduled task ... that's all that's required

Start cmd.exe as administrator

fsutil.exe behavior query disabledeletenotify

If you see "DisableDeleteNotify = 0" everything is fine.

 

[s|b]

Can solid state drives also slow down over time like traditional hard
drives due to fragmantation or some other reason that is specific to
solid state drives?

When I bought an SSD I did some reading and I can only confirm what Andy
says. Defragmenting was disabled under W7 and CrystamDiskInfo confirmed
that TRIM (and NCQ) are enabled.

Because I have an Intel SSD I can use Intel SSD Toolbox to check some
other settings and to weekly optimize using Trim functionality.

 

[charlie]

When I bought an SSD I did some reading and I can only confirm what Andy
says. Defragmenting was disabled under W7 and CrystamDiskInfo confirmed
that TRIM (and NCQ) are enabled.

Because I have an Intel SSD I can use Intel SSD Toolbox to check some
other settings and to weekly optimize using Trim functionality.

Samsung 840 SSDs also have a utility that supposedly takes care of such
things.

 

[Scott]

Samsung 840 SSDs also have a utility that supposedly takes care of such
things.

Samsung SSD Magician.

 

[Yousuf Khan]

Can solid state drives also slow down over time like traditional hard
drives due to fragmantation or some other reason that is specific to
solid state drives?

Not due to fragmentation, so don't ever run a defragmenter on it.

But SSD's use some reserved flash cells to speed up writing. As the SSD
gets filled up, these reserves start dwindling, especially in older,
smaller SSD's. So it takes longer to erase these cells before they can
be re-written on. So write speeds may go down, but read speeds should
not be affected at all.

Yousuf Khan

 

[cameo]

Can solid state drives also slow down over time like traditional hard
drives due to fragmantation or some other reason that is specific to
solid state drives?

RAM = Random Accessed Memory

Takes the same amount of time to access any memory address as another
or, conversely, doesn't take any more time to access one memory location
than another.

There is some tiny organizational overhead to manage the file system.
SSDs also incorporate wear levelling to maximize their lifespan; else,
reusing the same location for the same temporary data would result in
corrupted data when that block failed. Also, there is reserved space
(just like for hard disks) to map out bad blocks. Redirection due to
masking incurs a small overhead. None of these factor in regarding
fragmentation of files on a SSD. They are overhead for its normal
operation regardless of fragmentation.

http://en.wikipedia.org/wiki/Solid-state_drive
Fragmentation
There is limited benefit to reading data sequentially (beyond typical FS
block sizes, say 4kB), making fragmentation negligible for SSDs.[79]
Defragmentation would cause wear by making additional writes of the NAND
flash cells, which have a limited cycle life.[80][81]

For SSDs and regarding defragmentation: don't do it. No benefit but
generates further and unnecessary junction oxide stress (wear).

OK, thanks, though I always understood RAM to be the fast memory that
the CPU accesses directly, such DDR3, etc., not SSD.

 

[VanguardLH]

Can solid state drives also slow down over time like traditional hard
drives due to fragmantation or some other reason that is specific to
solid state drives?

RAM = Random Accessed Memory

Takes the same amount of time to access any memory address as another
or, conversely, doesn't take any more time to access one memory location
than another.

There is some tiny organizational overhead to manage the file system.
SSDs also incorporate wear levelling to maximize their lifespan; else,
reusing the same location for the same temporary data would result in
corrupted data when that block failed. Also, there is reserved space
(just like for hard disks) to map out bad blocks. Redirection due to
masking incurs a small overhead. None of these factor in regarding
fragmentation of files on a SSD. They are overhead for its normal
operation regardless of fragmentation.

http://en.wikipedia.org/wiki/Solid-state_drive
Fragmentation
There is limited benefit to reading data sequentially (beyond typical FS
block sizes, say 4kB), making fragmentation negligible for SSDs.[79]
Defragmentation would cause wear by making additional writes of the NAND
flash cells, which have a limited cycle life.[80][81]

For SSDs and regarding defragmentation: don't do it. No benefit but
generates further and unnecessary junction oxide stress (wear).

OK, thanks, though I always understood RAM to be the fast memory that
the CPU accesses directly, such DDR3, etc., not SSD.

SSD = Solid State Drive

http://en.wikipedia.org/wiki/Ssd
"uses integrated circuit assemblies as memory to store data
persistently"

Memory-based mass storage devices (aka drives) are not new. I remember
seeing some over 30 years ago but they were very low capacity because
memory was extremely pricey back then. They were also huge in size.

 

[Gene E. Bloch]

OK, thanks, though I always understood RAM to be the fast memory that
the CPU accesses directly, such DDR3, etc., not SSD.

RAM means Random Access Memory, i.e., memory that is access by directly
addressing a location. It is used as you say, but that is not the only
use.

PROMs, SSDs, thumb drives, SD cards, BIOS ROMs are a few uses of RAM.
Although people do not usually refer to some of those as RAM, they
really are.

On some level, even magnetic or optical disk drives are random access -
they just require a bit of maneuvering to get to the addressed
locations. But *nobody* ever calls them RAM

 

[athiker]

Can solid state drives also slow down over time like traditional hard
drives due to fragmantation or some other reason that is specific to
solid state drives?

When I bought my SSD from Crucial and discussed the issue with them
they also said not to defrag the disk.

They did say that about once a month I should start the system to the
BIOS and leave it overnight. "It will take care of itself."

Don't have a clue as to what "it" does.

 

[Paul]

When I bought my SSD from Crucial and discussed the issue with them
they also said not to defrag the disk.

They did say that about once a month I should start the system to the
BIOS and leave it overnight. "It will take care of itself."

Don't have a clue as to what "it" does.

The SSD has its own processor inside, and firmware.

The processor and firmware, run in the background, and rearrange
data to suit wear leveling and future performance.

The reference to "leave it overnight", applies to "4KB-sized torture tests".
The natural block size of SSDs is rather large. And they don't deal with small
files well. If you pummel the SSD with random small files like that (as Anandtech
does in some of its testing), the SSD drive needs significant time in the
background, to unravel the mess. The files are recorded instantly, the mess
is cleaned up later. The blocks would be moved around to consolidate space.
And it really does take all night.

If the drive is not allowed to do that, you could go from a drive that
has a 250MB/sec write rate, to dropping down to 100MB/sec. If you leave
the drive overnight, and powered up, the SSD processor and firmware
move those 4KB file fragments around. The drive then has the 250MB/sec
performance the next morning. Doing such movements, costs something
in terms of SSD lifetime.

More info here if you're curious. Or one of the many articles
on Anandtech probably explains it as well.

http://en.wikipedia.org/wiki/Write_amplification

The 4KB random file test, is a pathological one. If you're doing
activities like email or web surfing, the SSD likely isn't compromised
at all, and ten minutes of background work is enough to maintain it.
But if you use one of those special "torture test" programs, then
you'll need all night to restore full SSD write speed.

Other options for resetting an SSD, might include Secure Erase, which
is a feature of the ATA command set. That might be a quicker way to
straighten up an SSD that has received a torture test. But that
also erases the data.

There's plenty written about the care and feeding of SSDs out
there, but I don't have the time to read it all. It's worth
reading though, if you own one.

Paul

 

[Robin Bignall]

The reference to "leave it overnight", applies to "4KB-sized torture tests".
The natural block size of SSDs is rather large. And they don't deal withsmall
files well. If you pummel the SSD with random small files like that (as Anandtech
does in some of its testing), the SSD drive needs significant time in the
background, to unravel the mess. The files are recorded instantly, the mess
is cleaned up later. The blocks would be moved around to consolidate space.
And it really does take all night.

Thanks for yet another very informative post, Paul. Is it merely
necessary to leave the system switched on overnight, or does one
actually have to boot into BIOS and leave it there overnight. In other
words, is an idle Windows 7 idle enough to accomplish the task or not?

 

[Gene E. Bloch]

Thanks for yet another very informative post, Paul. Is it merely
necessary to leave the system switched on overnight, or does one
actually have to boot into BIOS and leave it there overnight. In other
words, is an idle Windows 7 idle enough to accomplish the task or not?

My guess is that since being in the BIOS avoids disk activity, it's not
gong to be at cross-purposes with the drives' firmware, whereas Windows
might do enough, even when ostensibly not being used, to confuse the
process.

My other guess is that the advice that athiker got is actually bogus.

Of course, in both cases, the operative phrase is "my guess".

 

[Paul]

My guess is that since being in the BIOS avoids disk activity, it's not
gong to be at cross-purposes with the drives' firmware, whereas Windows
might do enough, even when ostensibly not being used, to confuse the
process.

My other guess is that the advice that athiker got is actually bogus.

Of course, in both cases, the operative phrase is "my guess".

There's some difference between drives that support TRIM, and the
system is set up for TRIM. Some SSDs have the capability to tidy
up, even without TRIM, and those can run overnight. Just needs
power to work.

The difficult part, is figuring out what the drive is doing.
There is no need for a disk activity light, when the SSD drive
does stuff in the background. One way to determine that is going on,
would be to monitor the power consumption, which is a bit difficult.
A Molex to SATA power cable, a clamp-on ammeter, and you could monitor
for current flow. At some point, the current should drop back when the
drive is idle internally.

Paul

 

[DanS]

RAM means Random Access Memory, i.e., memory that is access by directly
addressing a location. It is used as you say, but that is not the only
use.

PROMs, SSDs, thumb drives, SD cards, BIOS ROMs are a few uses of RAM.
Although people do not usually refer to some of those as RAM, they
really are.

On some level, even magnetic or optical disk drives are random access -
they just require a bit of maneuvering to get to the addressed
locations. But *nobody* ever calls them RAM

Not necessarily true.......

en.wikipedia.org/wiki/DVD-RAM

 

[Gene E. Bloch]

Not necessarily true.......

en.wikipedia.org/wiki/DVD-RAM

Ah! I had forgotten about that - and I even have some and have played
with them

 

[Robin Bignall]

My guess is that since being in the BIOS avoids disk activity, it's not
gong to be at cross-purposes with the drives' firmware, whereas Windows
might do enough, even when ostensibly not being used, to confuse the
process.

My other guess is that the advice that athiker got is actually bogus.

Of course, in both cases, the operative phrase is "my guess".

I just had a look at Task Mgr / Performance with nothing running except
background tasks. I then exited most of those except virus checker.
After a few seconds it settles down to 0-1% CPU usage, with the idle
task mainly at 99%. That's probable even more idle than the average
British navvy, and should do.

 

[VanguardLH]

RAM means Random Access Memory, i.e., memory that is access by directly
addressing a location. It is used as you say, but that is not the only
use.

PROMs, SSDs, thumb drives, SD cards, BIOS ROMs are a few uses of RAM.
Although people do not usually refer to some of those as RAM, they
really are.

On some level, even magnetic or optical disk drives are random access -
they just require a bit of maneuvering to get to the addressed
locations. But *nobody* ever calls them RAM

The typical delineation is between "memory" and "mass storage subsystem"
to differentiate those types of devices. However, there once was
magnetic bubble memory that use magnetism as do hard disks to control
the dipolar alignment of the bubble "bits". Unlike memory of that day
that requires voltage to maintain the junction latch state, you could
completely remove power from bubble memory. Despite the much higher
price, HDD makers got scared so they upped their capacity to outstrip
and nullify the permanence of bubble memory.

http://en.wikipedia.org/wiki/Bubble_memory

As far as I remember, bubble memory was slow but it was non-destructive.
EEPROMs used for flash memory (USB thumb drives, SSDs) is destructive
due to oxide stress and why they incorporate wear levelling and bad
block remapping.

Although switching between cylinders which are groups of tracks to move
between sections of memory is something like switching between blocks of
memory (bits aren't accessed but blocks, like bytes or more likely
words), that's a bit of a stretch. Probably better delineation is one
based on how the device is physically fabricated: semiconductors with
dielectric charges versus dipolar alignment in magnetic media.
Differentiation is more about how they are made rather than on their
hardware interface.

 

[VanguardLH]

When I bought my SSD from Crucial and discussed the issue with them
they also said not to defrag the disk.

They did say that about once a month I should start the system to the
BIOS and leave it overnight. "It will take care of itself."

Don't have a clue as to what "it" does.

Many drives (SSDs and even HDDs) use their own firmware during idle
periods to retest any suspect sectors (or blocks of them). During
operation an error may occur during a read operation but later reads
pass. The sector becomes pending and gets tested later to see if it is
okay. Errors can be caused by gamma radiation, other hardware, or some
other glitch in the system beyond the device itself or even within the
device but it is not a critical or repeatable failure. I forget the
actual name of this feature but is is something a quiescent diagnosis
and correction procedure that is built into the device.