Follow up to defrag validity discussion: please read
in [Storage]
|
Prev: Samsung HD753lj external drive- computer does'nt see
Next: Hands-on with the ioSafe Solo external hard drive
From: David Brown on 10 Jan 2010 15:39 Arno wrote: > David Brown <david.brown(a)hesbynett.removethisbit.no> wrote: >> Cronos wrote: >>> David Brown wrote: >>> >>>> Nothing came close to being a sensible explanation of the claimed >>>> effects. >>>> >>> OK, I sent them your comments and am not going to post any more on the >>> subject. I will just have to find out for myself because there sure are >>> lots of people out there giving false info if what you say is correct, >>> an awful lot. > >> There are indeed lots of people, many claiming to be experts, giving >> false information on many subjects in IT - fragmentation is only one of >> them. > > All too true. And many want to sell you some product, which makes > it even worse. > If it is something like Diskkeepers spam, then at least it is honest spam - they don't deny that they are selling the software. The worst is the astroturfers - "I'm just a satisfied user", who are coincidentally posting from the company's network. >> Some of these are historical - fragmentation definitely used to be a >> major problem in the DOS/Windows world (*nix and other systems all >> handle fragmentation far better, and always have done). > > Interestingly, I recently stumbled over a case of a really > badly fragmented ext2 Linux filesystem or rather a bunch of > really fragmented files. This is so rare that there > is no current defragmenter, the respective project > was canceled AFAIK, because nobody needed it. The thing > that happened here is that some application created a large > sparse file and then filled it in in random places until full. > Defragging is still easy, just copy the file once and delete > the original. After scripting that, the problem is gone. > In fact there are "defragmenters" available for Linux. Some are for specific file systems, and many such projects have stagnated because there is no real use for them. Then there are several general defragmenters which, as you suggest, simply make new copies of the files and replace the old ones (the "mv" in Linux lets you do this atomically, with no disruption to read-only usage of the file). Since Linux does not fragment files unless there is good reason to, and even then minimizes it, this method is quite effective. On systems that support extents, such as ext4 or xfs, it is very effective - if you really want to go through the effort of defragmenting.
From: Rod Speed on 10 Jan 2010 15:56 David Brown wrote: > Cronos wrote: >> David Brown wrote: >> >>> Nothing came close to being a sensible explanation of the claimed >>> effects. >>> >> >> OK, I sent them your comments and am not going to post any more on >> the subject. I will just have to find out for myself because there >> sure are lots of people out there giving false info if what you say >> is correct, an awful lot. > > There are indeed lots of people, many claiming to be experts, giving > false information on many subjects in IT - fragmentation is only one > of them. > > Some of these are historical - fragmentation definitely used to be a > major problem in the DOS/Windows world (*nix and other systems all > handle fragmentation far better, and always have done). Some of it is > due to commercial forces - there are companies making a living out of > selling defragmentation software, and they need people to believe it > makes a difference. Sometimes it is due to mixing other effects ("I > cleared out my cache and temp directories, ran a registry tidier, ran > a spyware/adware remover, and defragged my disk - now the system runs > faster. Therefore defragging makes a difference"). > > Ultimately, only /you/ can say if defragging makes a difference for > /you/, on /your/ system, and the way /you/ use the computer. Thats overstating it. There are in fact very few personal desktop systems that benefit from defragging today. You can certainly make a case that defragging in the background when nothing else is being done does no harm, but its still pointless when it only benefits a few situations like file copying which is much better not done in the first place.
From: David Brown on 10 Jan 2010 17:55 Rod Speed wrote: > David Brown wrote: >> Cronos wrote: >>> David Brown wrote: >>> >>>> Nothing came close to being a sensible explanation of the claimed >>>> effects. >>>> >>> OK, I sent them your comments and am not going to post any more on >>> the subject. I will just have to find out for myself because there >>> sure are lots of people out there giving false info if what you say >>> is correct, an awful lot. >> There are indeed lots of people, many claiming to be experts, giving >> false information on many subjects in IT - fragmentation is only one >> of them. >> >> Some of these are historical - fragmentation definitely used to be a >> major problem in the DOS/Windows world (*nix and other systems all >> handle fragmentation far better, and always have done). Some of it is >> due to commercial forces - there are companies making a living out of >> selling defragmentation software, and they need people to believe it >> makes a difference. Sometimes it is due to mixing other effects ("I >> cleared out my cache and temp directories, ran a registry tidier, ran >> a spyware/adware remover, and defragged my disk - now the system runs >> faster. Therefore defragging makes a difference"). >> >> Ultimately, only /you/ can say if defragging makes a difference for >> /you/, on /your/ system, and the way /you/ use the computer. > > Thats overstating it. There are in fact very few personal > desktop systems that benefit from defragging today. > > You can certainly make a case that defragging in the background > when nothing else is being done does no harm, but its still pointless > when it only benefits a few situations like file copying which is much > better not done in the first place. > /I/ know that. But for people unconvinced by the sort of reasoning seen in this thread, doing their own testing and timing is the only way for them to see the reality of fragmentation.
From: Rod Speed on 10 Jan 2010 22:33 David Brown wrote: > Rod Speed wrote: >> David Brown wrote: >>> Cronos wrote: >>>> David Brown wrote: >>>> >>>>> Nothing came close to being a sensible explanation of the claimed >>>>> effects. >>>>> >>>> OK, I sent them your comments and am not going to post any more on >>>> the subject. I will just have to find out for myself because there >>>> sure are lots of people out there giving false info if what you say >>>> is correct, an awful lot. >>> There are indeed lots of people, many claiming to be experts, giving >>> false information on many subjects in IT - fragmentation is only one >>> of them. >>> >>> Some of these are historical - fragmentation definitely used to be a >>> major problem in the DOS/Windows world (*nix and other systems all >>> handle fragmentation far better, and always have done). Some of it >>> is due to commercial forces - there are companies making a living >>> out of selling defragmentation software, and they need people to >>> believe it makes a difference. Sometimes it is due to mixing other >>> effects ("I cleared out my cache and temp directories, ran a >>> registry tidier, ran a spyware/adware remover, and defragged my >>> disk - now the system runs faster. Therefore defragging makes a >>> difference"). Ultimately, only /you/ can say if defragging makes a difference for >>> /you/, on /your/ system, and the way /you/ use the computer. >> >> Thats overstating it. There are in fact very few personal >> desktop systems that benefit from defragging today. >> >> You can certainly make a case that defragging in the background >> when nothing else is being done does no harm, but its still pointless >> when it only benefits a few situations like file copying which is >> much better not done in the first place. >> > > /I/ know that. But for people unconvinced by the sort of reasoning > seen in this thread, doing their own testing and timing is the only > way for them to see the reality of fragmentation. Thats overstating it too. All but the silliest would likely be convinved by proper documented tests that someone else does and those that wont likely wont be convinced by their own tests anyway or even be capable of organising them, particularly with deliberately fragmenting a system. In other words it doesnt have to be on their own personal system.
From: David Brown on 11 Jan 2010 03:22
Rod Speed wrote: > David Brown wrote: >> Rod Speed wrote: >>> You can certainly make a case that defragging in the background >>> when nothing else is being done does no harm, but its still pointless >>> when it only benefits a few situations like file copying which is >>> much better not done in the first place. >>> >> /I/ know that. But for people unconvinced by the sort of reasoning >> seen in this thread, doing their own testing and timing is the only >> way for them to see the reality of fragmentation. > > Thats overstating it too. All but the silliest would likely be convinved by > proper documented tests that someone else does and those that wont > likely wont be convinced by their own tests anyway or even be capable > of organising them, particularly with deliberately fragmenting a system. > > In other words it doesnt have to be on their own personal system. > Cronos' big trouble, as I understand it, is that he is reading here that defragging has so little effect it is very seldom worth the effort, and elsewhere he is reading that it is absolutely essential. I'm not sure that /anyone/ is doing much in the way of "properly documented tests" - people like you and me are not interested in wasting the time and effort needed to produce a statistically valid result. However, it is worth noting that the people who really /do/ have an interest in producing such test results - the commercial defrag developers - have also failed to provide anything remotely like a properly documented and repeatable test setup showing the benefits of defragging. The fact that they have not produced anything other than numbers that seem to come out of a hat suggests that their numbers do, in fact, simply come out of a hat. Absence of proof may not be proof of absence, but in this case it comes very close. |