M1330 SSD and Disk Defrag = More harm than good?

I have had a M1330 now for about a week with a 64GB SSD. My concern about the drive is that since SSD functions differently than a HDD, will Disk Defrag cause more harm than good? I have Windows Live OneCare installed on my PC and has Tune-Up scheduled to run, though out of mild paranoia, I prevent it from running at times because after reading from various sources, some claim that it will help SSD and some say it will harm........so I'm a little bit lost here.....

 

Since every spot on the SSD accesses at exactly the same speed defrag shouldn't do much for you... On the otherhand doing a defrag will use up read/write cycles and reduce the lifetime of the SSD. So i wouldnt do Defrags, but thats just me.

 

i wouldn't too. because as the reason hawkshark stated...

 

Last i read the lifetime of the chips is greater than mechanical drives so throw out the window the eating up the HD all together. Studies need to be done to find out if is worth it.

 

I think those lifetime figures were only theoretical but 10 years of write/erase cycles at the least from what ive read

I saw a 32gb transcend SSD for sale around $300. Vista x64 doesnt even fit in there >=/

 

Regardless of wear leveling and failure statistics, it's still pointless on a(n) SSD and therefore a waste of your time...

 

I have a 64 GB SSD in my NC8430 and I've never defraged it.

 

Don't bother defragging - the whole purpose is to better your access times.

Since your access time is .1ms across the entire disk, it's a waste of time, and it re-writes the cells that have a limited life time anyway.

If its a good SSD it should outlast you, but like was said above it just isn't necessary.

 

Papi4baby is right, more testing really needs to be done. I have seen some data that suggests that while reads may not be severely affected by fragmentation; writes can be seriously slowed down. Dropping an 80MB/s operation to 10MB/s. So any file copies, installations, downloads could be affected. And since most SSD's are small right now fragmentation is more of an issue. Trying to pack Vista on a 64GB drive with your some basic software and it's already pretty full.

 

I have managed to stop OneCare from performing Disk Defrag by altering the regristry. Now I don't have to worry anymore


.....anyone here willing to put his / her SSD at risk for us so we can see some result (SSD vs Disk Defrag)?

 

its pointless because a SSD isnt a spinning disc, it has a 1ms access time all around so theres no need to sort out anything

 

..........LINK?

 

http://www.mwave.com.au/newAU/mwave...YMO2MPYWLTCCUZ9S6R8L3LSO1P5T253B&sku=22140016

Wow its actually gone down to $243AU thats about $210US

I dont think they deliver to US though

 

nvm apparently these transcend SSD's are very crap. I just read the NBR review for it

 

I found the 32Gb Transcend MLC 2.5" drive for IDE and SATA

The IDE One: ($170)
http://www.mydigitaldiscount.com/ProductDetail.jsp?LISTID=8000044A-1193346161

The SATA One: ($240)
http://store.sunwilltech.com/65342.html

These are the cheapest 2.5" ssd drives.

The reputation of these resellers are unknown, so you should check reseller ratings before purchasing.

K-TRON

 

Yea, i was surprised, for about a second....then i read

"Read up to 26MB/s, Write up to 13MB/s (32GB) "

And thats horrible, considering the newer SSD's get 100+mb/s read, and 80+write

 

It has been said already but let me say it a different way. Defragmenting a SSD is about as useful as tits on a bull! Unless Les says different! Because he can ask those who know.

 

Read the NBR review for transcend, its a horrible drive. No wonder its cheap. Must have been the first generation SSD's

 

It is also MLC, which means lower life span and slower speeds.

 

Transcend ssds aren't even worth considering. Don't look at the name of any ssd, but rather, the read/write speeds along with access times. Many learned right off with the slower Sandisk and Samsung what this would be like and complained after learning the hard way. The transcend is half that.

As far as defragmenting goes, it is not necessary because the ssd moves information around on its own as part of the wear levelling protection. You have no choice in the matter.

Unlike a hard drive which has all the information piled as close to the front as possible once defragmented, the ssd still spreads it all out evenly. Its a rather brilliant process. Even when you install another program, there will come a point when the ssd controller will rearrange data to best suit the drive itself.

As far as reaching the end of the ssd lifespan, yes it will be quicker on a mlc vice slc ssd but I on't really believe it something to concern yourself about.

 

I believe that you are right. I think it is for the reason described below but as I am not 100% sure that I got things right, I will have to disclaim myself .

I remember the data stored on an SSD is in data blocks. The data block size varies between different SSDs but I think I have seen figures around 64 - 128 kB. If you are writing 4 kB of data to an SSD, the SSD must rewrite the entire block i.e. if the block the SSD is writing to is partially filled with data, the SSD first needs to read the old data and then make a new write (old + new data) to the same block.

If your SSD gets fragmented in a way that leaves many of the data blocks partially full, then it is not strange that write performance of the SSD would degrade over time.

However, I am not sure if defragmenting the SSD would do you any good. Due to wear-levelling algorithms I believe that the internal data dispersal of an SSD does not match the layout presented over the disk interface. If that is correct, defragmenting your SSD is no guarantee for having the datablocks being fully filled or that your data is physically stored in one place on the SSD.

The expert on SSDs on Notebookreview is Les Tokar. Perhaps he knows if there are any benefits to defragmenting SSDs.

 

Ok so this affects the defrag argument in which way? I am confused? I thought defragging a SSD pushes the limits of common sense and strains the normal comprehension of reality? I mean just because SSD has the ability to set limitations why is that such a big issue? I know you will not live forever? Why your HDD has too, confuses me? WFT? Wear leveling just makes this an even more fun conversation!

 

jketzetera:

We responded at exactly the same time...see my post the page prior.

This argument seems to come up every month or so and always amuses me. Why don't we question hard drives in a similar fashion? When learning about ssds, I was afforded a white paper done by Google which dealt with the failure rate of hard drives which would surprise many.

As for the ssd, this is not new technology in the least. It is technology that has been used in the most critical of equipment for some time and is now leaking down to the consumer...which will be at a price to manufacturers if you believe it.

Where they once were able to charge exorbitant prices to enterprise for ssds, the force feeding of these at lower prices to the consumer (which is what will be occuring) will bring the overall industry price down.

Why would a company pay 8 dollars for a nut when that same nut costs a guy .58 cents at a hardware store?

 

I've been to plenty of blogs about this topic, and am grateful that some expertise exists in this thread. Thanks Les...

I also own an M 1330 and started a defrag, afterwhich I said D'Oh, and began searching for some answers. Having a Ph.D. in computer science I finally realized that I should get very little SPEED benefit except possibly when writing large files. However - I let the defrag run for a couple of hours and ended up with 5 gigabytes more space when it was done!!! Believe me, on a weenie 64 GByte drive using Vista, that is a whole world of difference.

1) I think I will run it in the future only when the disk gets within 15 percent of full,
2) use some discipline in keeping large images and videos that I don't really need on the drive, and
3) pray for a 128 GByte upgrade in the near future.

 

Thanks for the compliment Tom...

There is a 128Gb Memoright out if your interested. I think its around $3800 though here:

http://www.king-cart.com/cgi-bin/ca...emoright+2.5+128GB+SATA+SSD&exact_match=exact

eheh Its actually gone up since being 3200 a few weeks back. Can you imagine? An ssd truly for those that have money to burn!

 

Since Windows Indexing is enabled on my SSD, am I better off disabling it? Since SSD's have such a short access time?

 

I think for SSD's we may not be talking about traditional "defragmenting" and rather more like optimization to improve write performance. But without any hard data from testing on actual SSD machines we won't know. But if fragmented SSD's write at half the speed don't we want to find a way to get that performance back?

 

File fragmentation doesnt occur on ssds as it does with HDs. I also had never heard that fragmented ssds slow at all. Simply put, access speeds of the ssd are .1-.2ms and this does not slow over time.

We need to understand that the controller technology of the the ssd, for the most part, has a mind of its own. It moves data around on a constant basis where its best for 'wear levelling' without your knowledge. If you want to term this as defragmenting the ssd, so be it. Unlike a HD, the data does not remain in one spot which is the cause of much defragmentation over time with the HD.

Having said that, I will concede that using a defrag program that 'placs boot files first' did improve my startup time. I can't figure this out.

 

I have to say from my 1st impression of this thread I have been enlightened. I am intrigued by the disk space freed up most and now acknowledge more study/better explanations would be helpful. I am not willing to concede it is a good idea, as pointed out the wear leveling controllers determine where bits are placed. Not the OS or it's apps? I see no way around this? And while I do not see defragging as a major threat to the theoretical write limits I have concerns for doing unless demonstrated and explained why and how good. I expect more on this and Les as always I think you might provide insight, as many others here have also.

 

There are piles of articles that all relate quotes such as:
"It also means you've no more requirement to defragment your drive either" as can be found in this article by Bit-Tech.net and other reputable sources.

A simple search on Yahoo here will keep you up for a few nights reading much the same. Have your coffee ready!

At the end of the day, fragmentation of a hard disk slows the disk significantly whereas, this is not the case with a ssd. Its incredible access time of .1-.2ms leaves the HD miles behind even in its best shape and is the same from the first day you use it to the last.

From a very laymen view, I picture a hard disk which puts a piece of information permanently on the hard disk and spins at several thousand rpm causing fragmentation each and every time it accesses that. Parallel to that, I see a ssd which has a great index and directs itself to that same piece of information void of all the mechanical work and heat created. This same ssd has the ability to constantly monitor each and every block of storage (and there are literally too many to even consider counting) and move information accordingly to ensure they all wear evenly, thus adjusting the index and having the ability to still find that piece of info just as fast.

Very primitive yes... yet very advanced as it is wear levelling at its best.

 

I think that a lot of the "you don't need to defragment" comes originally from the manufacturers. And although I’m sure a heavily fragmented SSD will outperform any contiguous HDD, it seems that based on preliminary results there are some improvements to be had.

Because of the way cells have to be "erased" prior to a write function, it would seem to still be advantageous to have free space consolidated. Again, I don’t know that we’re really talking about “defragmentation” anymore than perhaps SSD optimization or more advanced wear-leveling. We should coin a new term and sell it to the OEMs!

Originally fragmentation wasn’t thought of as a problem until VAX guys found it was crippling their systems, and then FAT was supposed to help, and NTFS was supposed to “cure” fragmentation (which it didn’t). Now we are hearing about WinFS and SSD that are going to solve the problem for good. Just seems like this little bugger keeps coming back!

 

Flash memory does have negligible access times, but fragmentation could cause the filesystem drivers to slow down, as multiple accesses would have to be called. Even then, the possible improvements are not worth the shortened life of the drive.
With some modern filesystems, like ext3 and ReiserFS, fragmentation is such a small problem that it is generally not needed to defrag even for ordinary hard drives.

 

From a performance perspective, an SSD is not significantly different than RAM. A "sector" is an address and since this is a random access medium the response between two sectors at any addresses is essentially identical.

From an OS perspective, the controller gets a request for N number of blocks at various addresses. The device fulfills the request. The details are hidden. There are exceptions with things like NCQ (Native Command Queuing), but the basics don't really change.

 

I must admit that my low-level understanding of RAM is severely lacking. However, if you study the ATTO Diskbench graphs posted for various SSD:s, there is a clear pattern where random writing or reading small packets of data shows significantly worse performance than writing larger packets of data.

If you look at Les Tokars’s ATTO Diskbench Graph for Memoright’s 32GB SSD

http://forum.notebookreview.com/showthread.php?t=196354

you will notice that the sustained read/write speed when writing packets of 0.5 KB of data is less than 10MB/second. Increasing the packet size, increases the sustained write/read speed, until leveling off at about 64 KB packet size with approx 80/90 MB in sustained transfer speed.

It is my understanding that Windows uses blocks of 4kb data for the page file. Unless Windows’ disk caching algorithm (which I do not have a clue about) compensates for this, it would indicate that an SSD is not magnitudes better at accelerating page file access than a traditional hard drive, despite its magnitudes better access time.

 

Wow...thats a thought that I am waiting to here the answer on John and also, how would that compare to an HD then?

 

Any block I/O device does poorly with smaller I/O than larger I/O.

Correct, see above.

It's no worse for that either. Also these days with the amount of RAM on the machine, a machine with sufficient RAM should never swap. I've assumed here that the OS is written well too.

One caveat on "never paging" -- paging out is fine, as long as the page isn't recalled.

 

That's a flaw with the file system then; not with the device doing the I/O.

Exactly -- it's the nature of the file system more than the drive.

 

But that is a rather significant caveat. Anecdotally, I can tell you that Microsoft's operating systems just loooove the page file. They tend to page out memory to the page file, even if the memory load is less than 30% of available RAM (my personal experience). Just minimizing an application seems to represent a huge page-me-out flag to Windows 2000/XP.

Therefore, the question still stands. Does Windows caching algorithms compensate for the small page size (4kb) or do SSD's offer little advantage when it comes to page file access?

 

If I can just to stay along these lines... I am a speed freak with my system. I want to pull as much out of it as it will give which is why I went to things like ssd, 4Gb ram and 64 Bit.

I have my pagefil turned off and have for some time. I initially reduced it but soon learned that, even though its not needed, the system still uses pagefil as you just stated J.

It was only after I completely turned it off and forced everything to RAM that I observed yet again a speed increase. It wasn't significant mind you but it was visually noticeable. I really don't know of any other way to stretch my performance any further without possibly considering tapping with my system graphics but, in any case...

I have had nothing but positive results from having my pagefil turned off and the file itself deleted.

Heres my most recent PCMark05 score. It blows away pretty much every equal system as well as several with better CPUs and graphics. I am pretty proud of this and sure wish we could find some way of getting a better graphics adapter in this.