SSD Defrag mode

[vanuda]

Or actually not..

 

On my Thinkpad X300 ( early 2007 samsung ssd , no trim/ncq ) i can see a speed improve when doing a defrag..

 

Ok.. I know that this should not happen.. Normally the SSD will do

 

My guess here is that what i actually see is that the number of disc accesses due to directory/file fragmentation is minimized.

 

When booting the disc led is even rarely flashing..

 

I.e. there might be a need for a filesystem defragmentation ( not disc ) to improve SSD disc access..

 

When a file is defragmented in the filesystem and there are perhaps 100 fragments this slows down the access due to that the OS needs to do another 100 accesses to the disc...

 

Ok.. I know that this access is fast.. But if there is plenty of accesses like this this unnneccesarily slows down the access of a single file.. And if there is plenty of files this slows down the system..

 

True/False.. Cannot say.. But even if the access is fast excessive IO does punish the access of a file..

 

Maybe this kind of defrag should not be done more than perhaps 2-3 times per year..

[mta]

Firstly welcome!, but i'm sure you are about to get bombarded with repsonses, most negative, concerning SSD's and defragging.

 

it really boils down to a few simple points.

SSD's don't benefit from defragging.

Excess I/O should be avioded.

If you really want to - knock yourself out.

 

this topic gets raised a bit and the discussion is usually interesting

[Nergal]

Firstly welcome!, but i'm sure you are about to get bombarded with repsonses, most negative, concerning SSD's and defragging.

 

it really boils down to a few simple points.

SSD's don't benefit from defragging.

Excess I/O should be avioded.

If you really want to - knock yourself out.

 

this topic gets raised a bit and the discussion is usually interesting

what he said ;-)

[Alan_B]

On my Thinkpad X300 ( early 2007 samsung ssd , no trim/ncq ) i can see a speed improve when doing a defrag..

With all due respect, I feel that an observed speed improvement on a no trim/ncq SSD is not much more relevant to current technology than the physical layout of 80 column card punchers, sorters, and collators

 

The probability is that your SSD is thoroughly constipated because when a file is deleted the SSD flash is not disturbed - the controller merely remembers the space that can now be reused.

When a file is written it cannot be held in the space that can now be reused until the entire block containing that space has been erased,

and that will FIRSTLY require that UN-deleted files held in the same block must be relocated into other empty Flash.

 

Such an SSD system rapidly becomes very much slower when writing or updating files because when the user writes a 100 KB file,

there might be 100 MB of data that need rewriting from one block to another so that a complete block can be erased and start life again.

I know of this by the term "Write Multiplication".

A simple illustration,

Windows might write 10 GB to a new SSD and update those files 5 times before all 60 GB of Flash Cells have been written to.

After this for every 10 GB that is updated, 20 GB has to be written and shuffled around to erase complete blocks to which the new 10 GB update is written

 

I believe my SSD has a life of 10,000 Program Erase Cycles if I am lucky,

but on a user forum I was warned not to expect better than 3000.

I understand that more recent SSD with 3 bits per cell are much closer to single use and throw away

 

Always remember that an SSD has a life expectancy that is a bit better than the 9 lives of a cat.

A single pass defrag is likely to use up 3 Program Erase Cycle lives of every individual Flash cell,

and much worse if the free space was not contiguous and a preliminary pass was needed to obtain a large enough space.

[Alana]

So what your saying is, while SSD are fast, HDD are more durable? Go, HDD, go!!! LOL