I didn't "imply" anything. This is your excuse for another error on your part?
You implied Google switched to ext4. I've made no errors. In any case, I really doubt fragmentation is a concern for Google at all. Their BigTable database system would almost certainly have a very well tuned file allocation strategy which almost entirely takes the load off of the filesystem from making the decisions.
I merely stated the fact that Google use ext4 which has the properties that you were claiming could not co-exist. Why don't you tell Google their data is wrong?
Ext4 isn't really an example of a file system that puts incredible efforts into minimising fragmentation. It was a bad example on your part.
FYI what I said was: "
There do exist very specialist file systems that put incredible efforts into minimising fragmentation (some even simply don't allow it at all). But they lack in other areas, such as performance, scalability and hell even just reliability (non-journaling).".
I stand by that. Ext4 is a terrible example to provide to try to counteract that statement. It almost has nothing to do with what I was talking about, in fact. I was talking about specialist file systems - the sort of things they use on PVR set top boxes, and at the other extreme: the Mars Rovers. I didn't have a problem with you misunderstanding thinking I was taking a pot shot at Ext3/4, ZFS etc. But that was ultimately your mistake, not mine. Those filesystems are not what I'd consider specialist at all. A
specialist filesystem generally makes large compromises in certain areas in order to afford gains in others.
Of course it should have an online defragmenter, file fragmentation still exists.
Good. You finally concede it.
Please do not project your emotions onto the topic. Pointing out another factual error in your statement is not the same as "spinning" an advantage as a disadvantage (which clearly isn't possible).
What on earth? That would almost be funny if you weren't being serious.
The internet detective strikes! What next? statements about my mother? Please...
Again, what the hell?! This is getting pretty silly now, I hope your standard of posting picks up soon.
See this is why i'm replying, you are not aware. On ext3 or ext4 run fsck, for example. If we should be questioning anyone's credentials it should be yours
This is *nix for dummies, basic.
And how do you know that fsck is using the same criteria for detecting fragmentation that NTFS's standard defrag tool is? Hint: There are a lot of defraggers available for Windows, they will all return a different % figures for the same volume. Therefore differences between OS and filesystem are certainly going to exaggerate the subtle differences in the way that fragmentation is calculated.
So you don't have to prove your claim at all then? I have to disprove it? Brilliant logic there. What church indoctrinated you into this faith based thought?
Hmm, well no? Not really, no I don't. I'm not making outlandish claims that Ext3/4 perform better than NTFS or that they don't suffer from fragmentation to the same extent that NTFS does. All I've said was that NTFS is probably the most advanced file system available today. And it probably is once you take into account licensing, stability, fragmentation and scalability (in the context of all 3 key environments - server, desktop and workstation) concerns of the alternatives.
As I said before, you clearly have not used what you are trying to discuss. I work with extremely heavy datasets, I have a 9TB ext4 volume over LVM, I have not formatted for coming up to 2 years now. I have never defragmented it and it is sitting at 2% fragmentation. My home Windows 7 PC manages to get to 20-30% fragmentation within a month or two with less use. The difference is immediately apparent to anyone with even slight experience. Both frequently fluctuate between 10-40% free space. I'm using this as the example because you can test this at home and see for yourself.
But that's a stupid way to compare file systems: comparing NTFS running a desktop/workstation environment against a server storing large databases on Ext4. It's hardly any surprise NTFS scored less favourably in this scenario. A database server is highly optimised to avoid fragmentation and generally doesn't do lots of small file allocations like a typical desktop application does.
Unfortunately I don't have access to a "9TB ext4 volume" so no I can't "test this at home". But I fully expect that Ext4 would handle it fine, as would NTFS given a chance to handle the same workload.
Windows 7 defrags itself automatically (by default). So how you've got 20-30% fragmentation on it is quite some achievement. Not that Windows 7 reports a % figure anywhere regarding fragmentation - so I guess you just made it up.
In comparison to advancements in filesystems like ZFS (the context I used) it has.
You somehow managed to view massive overhauls as "constantly scratching their heads wondering and looking for a better file system for themselves. " creating != looking. It is just as easy to see gradual improvement as stagnation when stable examples in use on critical systems exist of a far superior filesystem.
ZFS is still missing pieces of its puzzle before it can be taken seriously for all environments (server, desktop, workstation). It needs a defragger.
Yeah it's terrible that people challenge you when you say something ridiculous with no basis and a refusal to back it up.
I don't mind discussion and even argument but the way you have gone about discussing this is completely over the top. That is why it is funny. Because I've seen it happen all too many times over the years.
Apparently I have to restrict my examples because you don't like them? Surprise! Because it doesn't fit your argument.
Please. ZFS has been around years as you have said yourself and I wouldn't call it's use in Solaris "Niché" considering the use it gets, especially in the US.
I've nothing against any of the filesystems you've suggested. (To do so would be ridiculous). They are all fair suggestions. However they don't really disprove my point that "NTFS is probably the most advanced file system available today". In fact, your suggestions have bolstered it because it has given me the chance to highlight some of their shortcomings.
ZFS has been around about about 5 years. And not all of those years was it stable or as feature-packed as it is today.
That was more a reference to the "ZFS is advanced, perhaps even on a par with NTFS" statement which is simply ridiculous for anyone with basic knowledge of NTFS / ZFS. Obvious trolling, nobody in the IT industry could be that oblivious.
Well that is your personal opinion. I'm only interested in discussing facts and logic here.
ZFS certainly has some killer features which NTFS does not. But still, it has other drawbacks that currently prevent it from hitting the big time. Until it gets a proper defragger tool, for example, it's never going to be a bit hitter on the desktop/workstation space (where fragmentation is often a problem).
The reason I agreed that ZFS is probably on a par with NTFS is because whilst it has certain advantages over NTFS (in terms of potential featureset), it also has some shortcomings (which I have pointed out elsewhere). Therefore, it seems fair to balance it by saying that the two are "on par". I don't think this is ridiculous and I think you would find most
true IT professionals would agree with me. Although I doubt any IT professionals exist that, through that profession alone, are qualified to compare filesystems in such detail. Yes they can look down a feature matrix and make a decision but they will glaze over pretty quickly if the subject of self-balancing tree data structures or delayed-write caching came up.
Please keep your personal vendetta against *nix users out of this. I use Windows too.
It's just an observation. You're not the only *nix user to have blindly assumed that NTFS is useless and that everything else is better. At one point you were even pushing ext3 above NTFS, though you seem to have conceded on those claims now, thankfully.
You certainly "fit the profile", as it were.
So you make a silly claim about NTFS. I provide examples which far surpass it in many ways and you don't like them because they don't fit your invisible criteria. You wouldn't need to do this if you had researched the topic first.
A claim which I have backed up numerous times with reasoning and logic. All you've done it banded about, in a rather chaotic fashion, the names of various alternative file systems. To which I have agreed are mostly good suggestions (especially ZFS, but definately not ext3) but which I still disagree are currently in a position to supercede NTFS as the "most advanced". Watch this space though.
This is pretty subjective in the case of ext4. I notice you omitted JFS / XFS.
Ext4 is lacking an online defragger. This prevents it from being a "serious contender" in the desktop/workstation space.
It's already fairly mature considering it's used for critical systems. It's pretty clear btrfs will replace it in the future; especially as Oracle bought SUN and intend to continue with btrfs development, though.
Yes it is mature, in the server space. But not in the desktop/workstation space.
It's understandable that the ZFS authors don't very much prioritise the development of a defragger - because their current user base simply don't need one for the usage scenarios it is currently being used for. And that's fine.
But until ZFS has proven itself more in the desktop/workstation space, it just isn't going to supercede NTFS in the minds of most. That is what gives NTFS the edge at the moment - it works well in all usage scenarios.
Yes, by saying "that's ridiculous here is why" and proving many of your statements to be riddled with a lack of knowledge and errors that makes me a troll.
Well, you have a very confrontational posting style and come across as though you're a bit hot under the collar. That much is obvious to most observers here. I didn't say the T word. You did.
) we are now slowly progressing out of this administrative model. We are not quite there yet but I'll expect this will finally change in the next couple of Windows releases. 
