Windows Vista: How bad is it to continue to use your pc during deep defragm
How bad is it to continue to use your pc during deep defragm
by Rammstein420 - 7/26/08 6:42 PMdefragmentation* i am using system mechanic to do a deep defrag (this takes days) the last time i did it and re analysed the system afterwards i found it was as defraged if not more.
Two things
by Jimmy Greystone - 7/26/08 8:37 PM In reply to: How bad is it to continue to use your pc during deep defragm by Rammstein420Two things:
1: Defragmenting really doesn't give any kind of a performance boost in probably 99% of common computer uses of the average person, and really is just unnecessary stress on the drive
2: If for some reason you're doing a lot of something that falls into that 1% exception area, then the more you use your system during a defragmentation session, the more likely files are going to be locked when the defragmenter wants to access them, because some other program is using them.
So basically, if you're not running a large database cluster, a high availability web server, or doing a lot of high end A/V editing, you're better off not even bothering with defragmenting. It will only offer meager and fleeting performance gains to programs where the performance is based on how fast the hard drive can retrieve data. As I said, that rules out at least 99% of all tasks commonly performed by your average user. For the overwhelming majority of us, it's a waste of time, effort, energy, and will send our hard drives to an early grave needlessly.
Of course I have an open challenge to anyone to find evidence to the contrary. So, if you can find a study done by some company that doesn't themselves SELL a defragmenting program, or that wasn't PAID for by a company that sells a defragmenting program... AND this study compares relative performance to things like web browsing, word processing, reading email, and listening to music files... I'd be interested in seeing it. I'm rather skeptical that any such studies even exist. I suppose it's possible a few were commissioned, but when no one could produce the results defragmenter peddlers wanted, they just buried them.
I slightly differ
by BigGuns149 - 8/4/08 12:15 AM In reply to: Two things by Jimmy Greystone"1: Defragmenting really doesn't give any kind of a performance boost in probably 99% of common computer uses of the average person, and really is just unnecessary stress on the drive"
99% is a bit of an exaggeration. While I agree that running a defrag utility isn't going to make huge across the board performance benefits, I would bet that in a non-contrived situation that one would notice a difference more than 1% of the time. Due to the nature of Bittorrent virtually anything that is downloaded with a Bittorrent client will be fragmented. Some files will be in hundreds of fragments to a dramatic enough degree that playback of video files will actually skip. If you think that downloaded video files off a torrent site is a contrived example than you don't realize how popular Bit torrent traffic really is online.
"So basically, if you're not running a large database cluster, a high availability web server, or doing a lot of high end A/V editing, you're better off not even bothering with defragmenting. It will only offer meager and fleeting performance gains to programs where the performance is based on how fast the hard drive can retrieve data. As I said, that rules out at least 99% of all tasks commonly performed by your average user. For the overwhelming majority of us, it's a waste of time, effort, energy, and will send our hard drives to an early grave needlessly."
With modern NTFS partitioned drives fragmentation isn't nearly as much of an issue as it was on older FAT16 or FAT32 formatted drives, but an occasional defrag isn't likely to send hard drives to an early grave. I have drives that 8+ years old that still work that I have defraged frequently for years.
Even common sense tells you that occasional defrags shouldn't harm drive life much if they aren't beneficial. Most of the wear and tear on a drive is spent seeking things. Fragmentation increase seek time, which means that it takes more time for the drive to spin down. Every second less that the drive is spinning is one less second that the motor is moving towards the grave. While frivolous use of defrag utilities no doubt shorten the life of the drive, most users will replace the drive long before the drive even approaches MTBF. I encounter a *lot* of average users who have HDDs that are 5-8 years old and still work because their computer is hardly ever on. Most of the HDDs I have seen that had a premature failure weren't people who were defragging the drive every other day. Most of them were people who had BitTorrent clients downloading data nearly around the clock so the drive almost nearly spun down. If defrag kills so many drives where are all the articles about people who killed HDDs left and right by defragging often? The evidence for this claim is sorely lacking.
"Of course I have an open challenge to anyone to find evidence to the contrary. So, if you can find a study done by some company that doesn't themselves SELL a defragmenting program, or that wasn't PAID for by a company that sells a defragmenting program... AND this study compares relative performance to things like web browsing, word processing, reading email, and listening to music files... I'd be interested in seeing it. I'm rather skeptical that any such studies even exist. I suppose it's possible a few were commissioned, but when no one could produce the results defragmenter peddlers wanted, they just buried them."
The notion that a study is flawed because of who funded it is dubious logic. Some of the dumbest advice I have ever heard on computers has come from people who are on no vendors' payroll directly or indirectly. Attack the message not the messenger. Furthermore, any study would also have to consider video playback of video downloaded off the internet, where defragging the the file that will be played back can certainly be noticed even in a double blind test.
Furthemore, why would Microsoft spend money to license a defrag utility if 99% of their users wouldn't benefit from it? Even if the licensing were a few cents per Windows license that would still easily be in the millions of dollars worth of lost profit.
Defragging one's computer in general won't result in dramatic performance gains, but I think you are marginalizing this oft tip a bit more than is justified.
Interesting
by Jimmy Greystone - 8/4/08 11:05 AM In reply to: I slightly differ by BigGuns149Interesting, a well reasoned and intelligent response. Usually all I get are a bunch of emotional knee-jerk reactions where try as you might, you can't figure out what the person's point was, assuming there was one. So, for that alone, I thank you.
Okay, you're right, 99% may be a bit of an exaggeration, but not by that much. And honestly, any decent bittorrent client will allocate the space for the file before it even begins downloading. I know Azureus/Vuze and uTorrent both do this.
Per my comment about sending drives to an early grave. Most of the people who would defragment way too often are people who wouldn't really know how to diagnose that it was their excessive defragmenting that helped kill it. Pretty much every staff reporter for Cnet, and almost the entire rest of the major computing rag outlets, clearly need a giant post-it note to remind them where the power button is on their computer. They probably have to use Wikipedia to figure out what a hard drive is.
I still personally wish Microsoft would "innovate" an idea that's been in place on Linux for several years now. Instead of using a simple FIFO queue for writing files, the operating system looks for the first available chunk of space that will fit the ENTIRE file. Failing that, it will break it up into as few pieces as possible. The result is an effectively self-defragmenting filesystem. On any Linux system I've ever run, even when I had large amounts of source files sitting around because I would compile absolute bleeding edge stuff (up to the hour) and use it, I never had fragmentation levels over 3%. One would think that such a "highly innovative" company like Microsoft, with its vast stockpiles of cash, could "innovate" some feature like this. Since "innovate" seems to have been co-opted by Microsoft to mean "copied" or "stolen", just like they've completely basterdized the word "genuine".
ANYWAY...
I think we can all agree that bad advice isn't hard to come by no matter where you look. The reason I gave the restriction of it not being paid for by some company that sells a disk defragmenter, is because if they're paying for it, they're going to want specific results. Namely that defragmenting is this absolutely wonderful performance enhancing thing that everyone should do. If they don't get those results, they'll likely shop around to look for a testing company that WILL give them the results they want, and all the other studies that show little to no gain will be buried and never see the light of day.
Microsoft's various anti-Linux studies are clear proof of this. And if it seems like I'm picking only on Linux, then you can look at how nVidia was gaming their drivers to detect benchmarking programs and then apply special settings that would give better results. A more recent example is how if you run a memory benchmark on Via's Nano CPU you get significantly different results depending on whether or not the Via CPU is identifying itself as a Via chip or an Intel chip.
And honestly, who in their right mind would put up the results of a study that show some product you sell to be of marginal usefulness at best?
As to why Microsoft licenses a defrag utility... Because they're hamstrung by their own commitment to backwards compatibility. It's also a good chunk of why Windows is still so incredibly insecure, tends to fall over at the slightest of nudges, and the Vista developers were bailing water like crazy, and throwing overboard everything that wasn't nailed down, just trying to keep that ship afloat. People expect there to be a defragmenter, so even though that era has long since passed, it sticks around. When Microsoft finally puts out a completely rewritten version of Windows, or even just retires Windows for something new, I wouldn't be at all surprised if the defragmentation tool goes away.
I'll just leave it with this little nugget... Old computer died after a severe manhandling by UPS when I moved. So, wind up with a new computer running Vista. Poking around looking for tips on improving performance, etc just to kind of kill time. Come across a suggestion to replace the Windows defragmenter with one from AusLogic or something like that. They also claimed to have a registry defragmenter, and I was intrigued by the notion. Anyway, on the defragmenter page, they mention how all computer experts agree that defragmenting, among other things, prevents program crashes. But you don't need to take my word for it, you can see a direct quote taken from here: http://www.auslogics.com/en/software/disk-defrag
"Disk fragmentation leads to system slowdowns, PC crashes, slow startups and shutdowns."
Now personally, I'd like to know how shuffling files around on the physical surface of the drive will do ANYTHING for "PC crashes".
There's also the bit a ways down on that page where they start off with something true, about hard drives being the slowest part of the computer, but then out of nowhere claim it's because they have moving parts. And that read/write head movements are what causes system freezes. Apparently they're banking on no one remembering that the defragmenting process involves significant amounts of read/write head movement... So, if we follow their line of reasoning, the process of defragmenting should be causing system freezes, but strangely it doesn't.
Anyway, I'd like to again thank you for the intelligent and well reasoned response. A very refreshing change. You get so used to the people who think that because they've mastered how to turn their computer on and launch 2-3 applications, it makes them some kind of expert, and that their opinion is the only one that should matter, you almost forget what it's like to have an intelligent discussion.
Defragging
by Ken Deyell - 8/5/08 8:50 AM In reply to: Interesting by Jimmy GreystoneI have read all these postings with interest,and I just wanted to mention that I am using both of the Auslogic's disk & registry programs.I have Vista a dual core AMD Athlon 4000+ processor,and there is a noted performance difference after their use.I do a lot of music file downloads and deletions so I use the disk defrag. prog. frequently.Highly recommend both programs they have been completely problem free.
Here's the real question
by Jimmy Greystone - 8/5/08 9:32 AM In reply to: Defragging by Ken DeyellHere's the real question though: Have you actually taken a stopwatch and timed various operations performed once after a reboot, and repeated the process several times to develop an adequate set of data points to draw a statistical conclusion; or did you just expect things to be faster, so whether or not they actually were, you convinced yourself that they were?
Because without some hard data -- collected in a way that should eliminate the effects of caching on part of the OS which would skew the results -- all you really have is, "It seems faster." Problem is, you don't really know if it is actually faster, or you just expected it to be faster, so you're paying more attention to things, and decide that it must have taken longer before. Whether or not it did is impossible to say, because no one ever has any empirical data to back it up.
And yes, ask any psychologist who researches cognitive functions, and they should be able to tell you without hesitation that the human mind's perception of time is extremely malleable. You don't even really need a psychologist for that matter. You've got little phrases in the lexicon like, "Time flies when you're having fun," to further back it up.
So what we have is some guy/woman who's just short of spamming these forums to promote his/her employer's product, and a couple of people who think things seem faster, but don't really have anything to back it up but a perception.
When you get down to it, more RAM and a bigger buffer on the hard drive will make world's more difference than all the defragging in the world. Because with platter sizes remaining fixed, and only the density with which data is stored changing, fragmentation levels become less and less of an issue. There's a maximum distance the read/write heads have to travel, and the more densely packed the data, the less likely they will ever have to travel that entire distance.
Now, if there were a defragmenter that also took into account the physics of angular velocity, and even more so, you could put some of the most commonly used files in the places of highest angular velocity, THAT might be worth having and using. Of course that would require deep inspection of every file on the drive, and would take forever and a day to complete. Not to mention it would likely be a real pain to write such a program.
Defragging
by Ken Deyell - 8/5/08 11:57 AM In reply to: Here's the real question by Jimmy GreystoneJimmy..your detailed critique of my comments I though was worthy of a reply.You are quite correct I do not have any hard facts to support my claim of a "performance improvement".However,even without the use of a stop watch one can guage the length of time it takes for a page to load when using Explorer.It has been my experience that,following a "defrag" the page will load in a flash rather then a pane at a time.This gives me a sense of improved performance,and crossing the finish line first.:)
Defrag
by barriepayne5 - 8/5/08 1:49 PM In reply to: Defragging by Ken DeyellTo Jimmy:
As someone who has been in the PC/Network administration field for over 25 years, I will second your comments 100%.
To Ken:
Judging the results of a hard disk defrag by browser loading times is inaccurate at best. A web pages data (unless cached) is coming from outside your computer. Even a super computer can load web pages slow if there is enough traffic slowing down a hosts upload speed.
The problem is
by Jimmy Greystone - 8/5/08 2:13 PM In reply to: Defragging by Ken DeyellThe problem is that the human mind is very susceptible to suggestion, even when you're the one doing the suggesting.
It's basically a case of the self-fulfilling prophecy. You go into something with a set of expectations, good or bad, and then you will shape your impressions around those expectations. So, if you go in thinking that defragmenting will provide all these wonderful performance benefits, your mind will be prone to fabricating these results in their empirical absence.
Yes, this DOES happen, and quite a bit more than you'd think. Human memory is NOT as accurate as we would like to believe. So you can't even really trust your perception of how long things took before defragmenting. If you choose not to believe me on that, you're free to go find out for yourself. Call up someone in the psychology department of a nearby community college or university. It may not be their area of expertise, but they should be able to confirm that there are numerous studies that all show this to be true. So unless I can somehow manipulate the answer you would receive from every psychologist in academia, they would have absolutely no reason to a) lie to you or b) confirm what I have said.
And, as pointed out, a web page is a horrible test for defragmenting performance gains. For one, you're downloading data which makes it immune to the immediate effects of fragmentation, and then there's a host of networking related reasons why loading a page two different times can have two very different results.
So, short of some empirical evidence in the form of timing specific events, no one's impression is worth squat. Not yours, not mine, not the Pope's, no one's. You can choose to believe it has some kind of benefit, but the truth is, you don't really know. Maybe it does, and maybe you're just deluding yourself into thinking it does.
Defrag Benchmarks Are Available
by Grif Thomas
- 8/5/08 4:08 PM
In reply to: Defragging by Ken Deyell
A little Googling finds quite a few benchmark references for you to compare with.. Although most benchmarking is done by third party companies which are promoting a product, they generally compare against "no defrag", or the "Windows Disc Defrag" tool, plus another competitor's product and generally, it still indicates that a badly defragged hard drive does get a performance boost by defragging..
Clicking on this one opens a PDF whitepaper regarding defrag stats on Windows Server 2003.
http://www.windows-tweaks.info/Downloads/O&ODefragV6_WhitePaper_EN.pdf
http://www.raxco.com/products/perfectdisk2k/performancetesting/PD7_DM3_Defragmentation_Performance_Test.pdf
http://donnedwards.openaccess.co.za/2007/11/benchmarks-diskeeper-2008-pro-premier.html
Hope this helps.
Grif
(NT) Grif ..as always you have the "source"
by Ken Deyell - 8/6/08 7:16 AM In reply to: Defrag Benchmarks Are Available by Grif Thomas
All three are invalid
by Jimmy Greystone - 8/6/08 11:05 AM In reply to: Defrag Benchmarks Are Available by Grif Thomas
All three of those are invalid, and here's why.
The first one gets a knock because it's a company trying to promote its own product. While they did provide far more details about how the testing was conducted and results collected, it's still a study done by a company looking to promote its own product. They're never going to publish some paper that says that the product they're trying to sell offers no real benefit. So that one's disqualified for conflict of interest.
The second one is simply invalid because they give no details on how the study was conducted or the results were collected.
The third is in the same boat as the second. No information how how the testing was done and results collected.
And before people start getting their panties all bunched, these aren't my arbitrary rules designed to create some impossible standard that no study could pass. These are all standard rules used by statisticians all over the world. How you conduct your testing and collect your results is usually far more important than the results themselves. I mean think about it. I could just claim that a survey of 10,000 people found that 95% hate Coca Cola products. Sounds pretty impressive doesn't it? However, how impressive does it sound when I then say that this survey was conducted at Pepsi headquarters, and only Pepsi employees were interviewed? What if I say that the Pepsi CEO sent out a memo before I arrived to conduct my survey, saying that anyone who said they don't hate Coca Cola products will be fired, and the CEO would be accompanying me personally while I did my interviews? Do these not seem like very important details? Details that might very well skew the results of the study.
Despite what it may seem like, my burden of proof is not really that high. There can't be an obvious conflict of interest (i.e. a company trying to promote its own product), there has to be information given about how the tests were done and the data was collected, and it has to relate to common desktop functions (i.e. web browsing, email, word processing, music playback). Now, assuming you can find a study where they actually bother to include info about how the tests were conducted and results collected, then we can at least say it passed the most basic level of scrutiny. Next will come digging into the details of how the testing was conducted and results collected, to look for any obvious flaws in the methodology.
reply
by barriepayne5 - 8/6/08 11:33 AM In reply to: All three are invalid by Jimmy GreystoneI olny looked through the second one but one thing that struck me was that the computer they were defragging had just short on 1 million defragmented files. This basically takes it out of the real world as most people don't have half that number of files on their whole computer much less just the dfragged ones.
You may have a hard time getting through. I have customers to this day who refuse to turn off their computers ever and who still want to run defrag once a week.
I read an impartial study once on defragging but for the life of me I can't find it again so this is just hearsay. The study showed that a pc with a 10% defragmentation could expect a 1-2% performance increase by defragging. it takes a while to get up to 10%. I haven't defragged my pc in over a year ad its at 13%. Just upping your pc from 512 to 1 or 2gigs of ram will give you way more performance increase than that.
I usually tell my customers that once a year is fine. Twice if they are heavy users and never is fine too if they don't want to.
I'd buy those results
by Jimmy Greystone - 8/6/08 1:26 PM In reply to: reply by barriepayne5I'd buy those results as being credible on the face of it. Just the usual disclaimer about that being subject to change should I ever actually see the study for myself.
And it is sad, I know. I was recently at a job interview where the person was asking me what I would do if someone was complaining of a slow system. After going through a fairly exhaustive list, they asked me if I'd consider defragmenting, and I told them that I've never found it to be of any real help, but if the user really wanted it, I'd run it for them. I didn't get the job, but more because I later found out they changed the job description to something completely different from what I applied for, and just neglected to tell me. I have no real qualms about mentioning this was Salesforce.com I interviewed with. I could even give specific names, but I'll just leave it at that I was left rather unimpressed with their operations, and a little annoyed that I spent quite a bit of money to fly in specifically for that interview just to be jerked around like that. Fortunately I had a friend who lived in the general area whose couch I could crash on.
ANYWAY, I think that's kind of the key here... While the hard drive is generally a performance bottleneck, the better solution really is to increase the amount of RAM available for caching frequently used data from the drive, as well as increasing the buffer size on drives. When the first 8MB buffer drives started coming out, you could see a noticeable improvement in disk throughput over older 2MB buffer drives. I don't think it's any real coincidence that there are now 16MB and even 32MB buffer drives.
I keep hoping solid state drives will really come of age soon. They're inching ever closer, but still not quite there yet. Then maybe we can create a new hard drive interface that can keep pace with the FSB a little better.
Roland, Interesting.. Different Figures Here..
by Grif Thomas - 8/6/08 4:26 PM
In reply to: reply by barriepayne5
Not that I disagree with your overall premise because many folks defrag way too often. This is certainly not a scientific test, but I just did a quick check on four XP PRO SP3 machines here by running the standard XP Defrag program in "Analyze" mode.. Since I perform the maintenance on all our computers and I KNOW that I performed a defrag on all 50+ machines last July 9 and 10, it is interesting to note that 3 of 4 have more than your 13% indicated, at least as it relates to "File Fragmentation". (Fragmenting "free space" has always been an interesting issue but I'll leave that issue aside.) It's only been about 4 weeks since a defrag was performed... See the results below:
Computer 1:
Total Fragmentation: 10%
File Fragmentation: 19 %
Computer 2:
Total Fragmentation: 13%
File Fragmentation: 26%
Computer 3:
Total Fragmentation: 1%
File Fragmentation: 2%
Computer 4:
Total Fragmentation: 10%
File Fragmentation: 20%
I defrag every couple of months on our work machines.. About the same on my home computer as well although the fragmentation percentage is generally higher on my home machine because of the things I do with it. And sorry about your RAM addition solution. I don't think that really applies. Hard drive performance still degrades from fragmentation, whether you not you add RAM.. Besides, extra RAM costs money.. Defrag doesn't..
Hope this helps.
Grif
| Forum legend: | |
 |
Locked thread |
 |
Moderator |
 |
CNET staff |
 |
Samsung staff |
 |
Norton Authorized Support team |
 |
AVG staff |
 |
Windows Outreach team |
 |
Dell staff |
 |
Intel staff |
