Sign in with
Sign up | Sign in
Your question

Is Defragmenting worth it?

Last response: in Storage
Share
April 28, 2012 12:05:58 AM

I probably haven't done a defrag on a computer in 7+ years. Even on Win XP I didn't do it often.

Haven't done it at all on my last 3 computers, not once on Windows 7.

Tell me why it's worth it. My Dad use to do it all the time. I remember he always said don't playing with the computer it's defragging! :D 

More about : defragmenting worth

April 28, 2012 12:15:57 AM

it depends, if you write/delete often too many files, then... yes.
if you don't, then you probably don't needed it.
m
0
l
April 28, 2012 12:20:07 AM

Back in the days when hard drives used to have capacities of 30GB, defragging had a large effect. Nowadays, not so much.

I've got a 300GB HDD from 2006. The last time I defragged it was a couple of days ago, and the time before that was probably at least 4 months ago. Anyway, I noticed no improvement in performance after the defrag.
m
0
l
Related resources
April 28, 2012 5:23:53 AM

just seems like a waste of time. hd's are huge and it would take forever.

I use a program called revo uninstaller and CC cleaner. Those are pretty good at keep everything organized.

m
0
l
April 28, 2012 7:31:56 AM

Nothing is worth anything unless it is called for. Analyse your hdd using the defragmentation tool. If a lot of red lines or large area of red is present, then defragmentation is definitely worth some of your time. If blue dominates, then probably not.
m
0
l
April 28, 2012 5:32:43 PM

zyzz said:
I probably haven't done a defrag on a computer in 7+ years. Even on Win XP I didn't do it often.

Haven't done it at all on my last 3 computers, not once on Windows 7.

Tell me why it's worth it. My Dad use to do it all the time. I remember he always said don't playing with the computer it's defragging! :D 


Defragging a HDD is certainly a good thing to do occasionally, especially if you do a lot of internet browsing, downloading and gaming. All those files including your OS files get fragmented over time. Windows takes milliseconds longer to find them. That being said do it occasionally for your hard disks, never on an SSD.
If your time is that valuable and your that impatient do it when you go to bed. Will be finished in the morning
m
0
l
April 28, 2012 5:34:43 PM

zyzz said:
just seems like a waste of time. hd's are huge and it would take forever.

I use a program called revo uninstaller and CC cleaner. Those are pretty good at keep everything organized.


....and your comment has what to to with "defragmenting".
m
0
l
April 29, 2012 7:24:40 PM

gerry410 said:
....and your comment has what to to with "defragmenting".

Defragmenting is about keeping files organized, together, and freeing up room. Am I wrong?

I just meant that I properly uninstall programs so there is nothing left. Maximum free space, organized, no temp stuff.

m
0
l
a c 353 G Storage
April 29, 2012 7:59:50 PM

Defrag is for one purpuse and that is to organize your files in a sequencial manner.
If you notice that for both HDDs and SSDs the Highest preformance is with Sequencial reads writes.

If you have a two partitions set up on your HDD, 1 For OS + programs (Your C drive, and the recommneded size for say a 1 TB drive would be around 300 MB) and a "D" drive for all of the files that you generate. MY PREFERED Method for several reasons.

For the OS +program drive. Three things cause: A) a swap file that is expanding and shrinking (PS I always set Min and max size to the same value), B) Temp files that get stuck on C drive and then deleted, and C) files that you delete. Normally only large files get fragmented: here is an example say you have a "small" 10 meg file. This file would require 10 megs space / 4KB ( typical cluster size ) = 2,500 clusters that ideally would be placed in Sequencial clusters and read at Sequencial speed, But if these same clusters are spread all over the HDD then there is a xx MSec delay as the HDD has to reposition the head all over the HDD. However Most files on an OS + Program drive are under 1 Meg (typically 50% of the Files are 16K and smaller. So Not as big of an effect.

For Your Files, if you are often creating and deleting large files a fragmented drive can become really boged down. I have single files that are 1 Gig (DVD dot VOBs) and 13 -> 40 gigs for a Blue-ray file - Heaven help reading that if it was highly fragmented.

Unless you have turned windows defrag OFF, it should atomaticalyy run in the back ground - reason most do not notice much def when they do a defrag - the drive was not that fragmented. For us "old" folks defrag had to be run manually, plus you had more control over where files and directories were stored and Mad one heck of a difference on them Old 20-> 40 MEG MFM and RLL Hard drives.

Bottom Line if you turn Defrag OFF, Yes you should periodicaly run it!
Added - DO NOT run Defrag on an SSD!! Not only will it Not perform noticably faster ( Access time is so much faster than a HDD), it has an adverse effect on it.

PS defrag does not really free up room.
m
0
l
a c 379 G Storage
April 30, 2012 3:19:41 PM

zyzz said:
Defragmenting is about keeping files organized, together, and freeing up room. Am I wrong?

I just meant that I properly uninstall programs so there is nothing left. Maximum free space, organized, no temp stuff.


Even if you properly uninstall a program, it now leaves a blank spot where the program used to exist. The next time you install a program, if it doesn't fit in that space, part of the program may end up there, and the rest of the program will end up somewhere else, thus fragmented. Fragmentation occurs quite a bit, especially as temp files get deleted and replaced. Installing, uninstalling, reinstalling often also causes fragmentation.

The point is fragmentation is natural and at some point the drive should be defragged.
m
0
l
a b G Storage
April 30, 2012 3:28:42 PM

Good post by RetiredChief.
The short of it all, since Vista, Defragment ran on a scheduler, in the background. Set it and forget it. Makes HDD happy, maintains performance, and you don't even realize it.

And to repeat NEVER run defragment on an SSD.
m
0
l
April 30, 2012 11:07:26 PM

got a major question here why is it that only windows had problems with fragmentation? in all my years i have never seen a mac or linux defrag. EVER.
m
0
l
a b G Storage
April 30, 2012 11:31:05 PM

pckitty4427 said:
Back in the days when hard drives used to have capacities of 30GB, defragging had a large effect. Nowadays, not so much.

I've got a 300GB HDD from 2006. The last time I defragged it was a couple of days ago, and the time before that was probably at least 4 months ago. Anyway, I noticed no improvement in performance after the defrag.


Just because you did not notice the performance doesn't mean you can scoff at defragmentation. Perhaps your HDD was minimally fragmented to begin with due to minimal usage. Or your defragmentation software simply sucks.

tbillion said:
got a major question here why is it that only windows had problems with fragmentation? in all my years i have never seen a mac or linux defrag. EVER.


All about filesystem. NTFS has a different way of file organization than EXT4 and whatever file system mac uses.
m
0
l
a b G Storage
May 1, 2012 9:49:53 AM

Quote:
tbillion wrote :

got a major question here why is it that only windows had problems with fragmentation? in all my years i have never seen a mac or linux defrag. EVER.


All about filesystem. NTFS has a different way of file organization than EXT4 and whatever file system mac uses.


It's the inherent nature of a rotating media (aka a Hard Disk). The HDD wants data to stay as close to the center as possible for performance reasons. As files (or blocks) are freed up near the center, the HDD wants to reuse them. However, sometimes that new file doesn't fit, so it breaks the file into pieces.

There has always been fragmentation of data and there always will be, but fragmentation of files/data is only a problem when you have physical mechanics at play (e.g., architecture of a record player). In solid state media (NAND/SSD, NOR, DRAM) fragmentation is actually a good thing, as the IO controller can effectively read and write as we understand RAID to function today.

(an eye opener, simply google "Defragment EXT4" - all kinds of instructions on how to defragment that filesystem)
m
0
l
a c 353 G Storage
May 1, 2012 12:23:46 PM

psaus - "The HDD wants data to stay as close to the center as possible" - Not really

Data Starts at the out circumference and works it's way in. Reason: The Highest performance is at the Outer edge of a platter, and the Lowest at the center - Performance in the middle is inbetween. Has to do with angular velocity, that is the time it takes to move say x number of clusters is Much higher at the outer edge than for the same number of Clusters at the center.. NOTE: the Outer edge is where you will find your Operating system.
m
0
l
a b G Storage
May 1, 2012 12:27:41 PM

^ you're completely right. *facepalm* one of my finer dyslexic moments. I'm very aware outter is faster, I've known this for years... can't believe I typed that.
The rest of my post is correct tho.
m
0
l
a b G Storage
May 1, 2012 8:02:06 PM

psaus said:
Quote:
tbillion wrote :

got a major question here why is it that only windows had problems with fragmentation? in all my years i have never seen a mac or linux defrag. EVER.


All about filesystem. NTFS has a different way of file organization than EXT4 and whatever file system mac uses.


It's the inherent nature of a rotating media (aka a Hard Disk). The HDD wants data to stay as close to the center as possible for performance reasons. As files (or blocks) are freed up near the center, the HDD wants to reuse them. However, sometimes that new file doesn't fit, so it breaks the file into pieces.

There has always been fragmentation of data and there always will be, but fragmentation of files/data is only a problem when you have physical mechanics at play (e.g., architecture of a record player). In solid state media (NAND/SSD, NOR, DRAM) fragmentation is actually a good thing, as the IO controller can effectively read and write as we understand RAID to function today.

(an eye opener, simply google "Defragment EXT4" - all kinds of instructions on how to defragment that filesystem)


I disagree.

Your Hard Drive Platter is a dumb piece of metal that is under Operating System/File System control. NTFS by its design does not care how files are organized. Think of your data as a library. Then think of your filesystem as a librarian. NTFS is a bad librarian while EXT4 is a better librarian. NTFS will take all the returned books from customers (your operating system/applications) and simply place them wherever that librarian finds a random open spot. Then when your customers (your OS/Apps) start looking for that book...its harder to find it, because its misplaced. EXT4 by its design is a better librarian by which is organizes the books differently and better for certain purposes.

http://geekblog.oneandoneis2.org/index.php/2006/08/17/w...

Secondly, fragmentation on an SSD is not a "good" thing or "bad" thing. It simply does not matter. All the cells are accessed at the same rate and time, therefore it does not matter if it is fragmented or not, real world performance will be the same. In addition, RAID with SSD's does not improve access times. It only helps sequential read and write speeds as far as I know (which can be achieved extremely high if you have enough HDD's). An SSD's access time is so miniscule that the overhead of a raid controller and two SSD's which just slow down access times.

Real World performance from SSD's do not come from the huge read/write speeds. It comes from the access times which allows the SSD to locate and extract files faster. Majority of OS's and Applications are made up of very small files. SSD's are amazing, because the way the file system organizes files or the OS does simply makes no difference. On a HDD, defragmentation lowers the latency that files are accessed due to them being close together and the reading head moves much less to find it.
m
0
l
a b G Storage
May 2, 2012 9:38:50 AM

@blackhawk192 - unfortunately it's clear you're a Linux supporter, as well that you don't have a full understanding of the underlying technology in today's x86/64 systems.
Sorry, do be a Linux supporter, I am too, there's nothing wrong with that. :)  But understand thoroughly before trying to defend your beloved OS.

http://ubuntuforums.org/showthread.php?t=1368737
http://forums.fedoraforum.org/showthread.php?t=232773

^2 very common builds of Linux talking about steps in how to defragment your EXT4 OS volume. You are absolutely correct that EXT4 prevents fragmentation much better than NTFS (as you say, "EXT4 is a better librarian). But it's the nature of the beast (HDD) that fragmentation is going to exist and hinder performance, no matter what file system you place on that antiquated medium (HDD).

Quote:
Secondly, fragmentation on an SSD is not a "good" thing or "bad" thing. It simply does not matter. All the cells are accessed at the same rate and time

This is how I could tell you don't understand the underlying technology - SSD Fragmentation is done by design in the controller (same with DRAM and its counterpart the memory controller). I wasn't terribly clear in my earlier post, but will clarify here. All NAND chips in an SSD are absolutely NOT accessed at the same time, the controller isn't that powerful (yet). The very first SSDs were 4 channels, today 8/10 channels are common in the fastest of SSDs, with 16 being seen in about ~12mo from now. All of these channels can be accessed simultaneously, but each channel has 2 - 8 NAND chips daisy chained to it in order to achieve the 100's of GB seen today as well as give you the throughput to now out pace HDDs (your higher end drives are using 128Gbit (16GB) NAND devices (more than 1 die)... e.g., a 320GB drive will have at least 20 NAND devices on it - but most likely there will be extra GB for "wearleveling overhead"). Intel's 200GB 710 torn down. You can think of these channels as a quazi-RAID, as the controller is going to break-up files into pages in order to stripe them across all available channels. But when the controller needs to switch between chip A and chip B in a given channel you will see a couple ns lag. However this lag pales in comparison to the several ms of lag a HDD experiences when it needs to shift locations. This principle applies when talking about DRAM modules, and why it's always advised to only populate the first socket of each channel, and populating the second & third socket of the channels reduces performance ever so slightly.
My point, DRAM and NAND controllers split pages across different chips (fragmentation), in a RAID-0-esque fashion to increase/optimize performance. If this didn't happen, your SSD might pull off a whopping 30-50MB/s R/W. :)  So, fragmentation, in it's basic sense, is a good thing for everything except HDDs.

Quote:
In addition, RAID with SSD's does not improve access times. It only helps sequential read and write speeds as far as I know (which can be achieved extremely high if you have enough HDD's). An SSD's access time is so miniscule that the overhead of a raid controller and two SSD's which just slow down access times.

Correct and incorrect.
Correct: 2 drives in a RAID result in minuscule performance increases. But this statement applies to HDD and SSD alike. Most performance gains in a 2 drive HDD RAID is the access time, but those gains are questionable in it's worthiness given all the headaches associated with RAIDs.
Incorrect: Because SSDs and HDDs scale similarly in RAID environments. This is demonstrated in enterprise grade tiered storage solutions which explicitly use SSDs for a newly created "Tier 0". It's access and throughput performance scales incredibly when you throw a whole mess of SSDs in a RAID10/5/50/6/60. The scaleability trend of SSD mimics that of HDDs in RAID configurations, but the end results of an SSD vs HDD resembles Tesla motors racing a Model-T in the Nuremberg Trials.
m
0
l
a b G Storage
May 2, 2012 7:53:13 PM

psaus said:
@blackhawk192 - unfortunately it's clear you're a Linux supporter, as well that you don't have a full understanding of the underlying technology in today's x86/64 systems.
Sorry, do be a Linux supporter, I am too, there's nothing wrong with that. :)  But understand thoroughly before trying to defend your beloved OS.

http://ubuntuforums.org/showthread.php?t=1368737
http://forums.fedoraforum.org/showthread.php?t=232773

^2 very common builds of Linux talking about steps in how to defragment your EXT4 OS volume. You are absolutely correct that EXT4 prevents fragmentation much better than NTFS (as you say, "EXT4 is a better librarian). But it's the nature of the beast (HDD) that fragmentation is going to exist and hinder performance, no matter what file system you place on that antiquated medium (HDD).

Quote:
Secondly, fragmentation on an SSD is not a "good" thing or "bad" thing. It simply does not matter. All the cells are accessed at the same rate and time

This is how I could tell you don't understand the underlying technology - SSD Fragmentation is done by design in the controller (same with DRAM and its counterpart the memory controller). I wasn't terribly clear in my earlier post, but will clarify here. All NAND chips in an SSD are absolutely NOT accessed at the same time, the controller isn't that powerful (yet). The very first SSDs were 4 channels, today 8/10 channels are common in the fastest of SSDs, with 16 being seen in about ~12mo from now. All of these channels can be accessed simultaneously, but each channel has 2 - 8 NAND chips daisy chained to it in order to achieve the 100's of GB seen today as well as give you the throughput to now out pace HDDs (your higher end drives are using 128Gbit (16GB) NAND devices (more than 1 die)... e.g., a 320GB drive will have at least 20 NAND devices on it - but most likely there will be extra GB for "wearleveling overhead"). Intel's 200GB 710 torn down. You can think of these channels as a quazi-RAID, as the controller is going to break-up files into pages in order to stripe them across all available channels. But when the controller needs to switch between chip A and chip B in a given channel you will see a couple ns lag. However this lag pales in comparison to the several ms of lag a HDD experiences when it needs to shift locations. This principle applies when talking about DRAM modules, and why it's always advised to only populate the first socket of each channel, and populating the second & third socket of the channels reduces performance ever so slightly.
My point, DRAM and NAND controllers split pages across different chips (fragmentation), in a RAID-0-esque fashion to increase/optimize performance. If this didn't happen, your SSD might pull off a whopping 30-50MB/s R/W. :)  So, fragmentation, in it's basic sense, is a good thing for everything except HDDs.

Quote:
In addition, RAID with SSD's does not improve access times. It only helps sequential read and write speeds as far as I know (which can be achieved extremely high if you have enough HDD's). An SSD's access time is so miniscule that the overhead of a raid controller and two SSD's which just slow down access times.

Correct and incorrect.
Correct: 2 drives in a RAID result in minuscule performance increases. But this statement applies to HDD and SSD alike. Most performance gains in a 2 drive HDD RAID is the access time, but those gains are questionable in it's worthiness given all the headaches associated with RAIDs.
Incorrect: Because SSDs and HDDs scale similarly in RAID environments. This is demonstrated in enterprise grade tiered storage solutions which explicitly use SSDs for a newly created "Tier 0". It's access and throughput performance scales incredibly when you throw a whole mess of SSDs in a RAID10/5/50/6/60. The scaleability trend of SSD mimics that of HDDs in RAID configurations, but the end results of an SSD vs HDD resembles Tesla motors racing a Model-T in the Nuremberg Trials.



LOLOL! I'm a Linux Supporter?...I am actually not. I have no linux computers or OS in my house. In-fact I hate Android. I use strictly Windows and Apple Operating Systems. I'm not defending anything, I'm simply showing you what the Internet Says about HDD's, File SYstems, and Defragmentation.

Quote:

This is how I could tell you don't understand the underlying technology - SSD Fragmentation is done by design in the controller (same with DRAM and its counterpart the memory controller). I wasn't terribly clear in my earlier post, but will clarify here. All NAND chips in an SSD are absolutely NOT accessed at the same time, the controller isn't that powerful (yet). The very first SSDs were 4 channels, today 8/10 channels are common in the fastest of SSDs, with 16 being seen in about ~12mo from now. All of these channels can be accessed simultaneously, but each channel has 2 - 8 NAND chips daisy chained to it in order to achieve the 100's of GB seen today as well as give you the throughput to now out pace HDDs (your higher end drives are using 128Gbit (16GB) NAND devices (more than 1 die)... e.g., a 320GB drive will have at least 20 NAND devices on it - but most likely there will be extra GB for "wearleveling overhead"). Intel's 200GB 710 torn down. You can think of these channels as a quazi-RAID, as the controller is going to break-up files into pages in order to stripe them across all available channels. But when the controller needs to switch between chip A and chip B in a given channel you will see a couple ns lag. However this lag pales in comparison to the several ms of lag a HDD experiences when it needs to shift locations. This principle applies when talking about DRAM modules, and why it's always advised to only populate the first socket of each channel, and populating the second & third socket of the channels reduces performance ever so slightly.
My point, DRAM and NAND controllers split pages across different chips (fragmentation), in a RAID-0-esque fashion to increase/optimize performance. If this didn't happen, your SSD might pull off a whopping 30-50MB/s R/W. :)  So, fragmentation, in it's basic sense, is a good thing for everything except HDDs.


Way to literal, its clear you have a mis-understanding of what I said/meant. The OP was talking about regular operating system fragmention and de-fragmention as an end-user. Wear Leveling controllers and other really fundamental levels of software controlling hardware surely have their processes that change a lot. However the OP was asking from a sense of de-fragmentation software on a computer. As an end-user, when you "defragment" your SSD, it doesn't matter because as far as you are concerned your performance will be the same, what looks to be like defragmented on the picture really might be mumbo-jumbo as the SSD controller/firmware organize everything differently anyway. You might be correct about how the technology works on a fundamental level...but you are explaining it way to specifically and unnecessarily. I meant fragmentation as the same meaning as the OP did which was software-based user fragmentation of files, not what happens physically to files inside the SSD LOL. I was correct as far as the defragmentation goes and you were also. Just a mis-understanding. :)  On terms with the RAID, I wasn't sure which is why I stated clearly "As far as I know". Covered my bases.

Your explanation is like explaining the mechanics and physics of how a clock works when all somebody did was ask for the time.
m
0
l
a b G Storage
May 2, 2012 9:26:47 PM

Quote:
Your explanation is like explaining the mechanics and physics of how a clock works when all somebody did was ask for the time.


Yea, off topic, but your "i disagree" comments and similarly detailed links prompted the need for clarification.
m
0
l
!