Tom Rizzo launches his new colum with an overview of why the new “Longhorn” storage subsystem (code-named “WinFS”) is needed, what WinFS promises to do to help solve our data-overload problems, and what his column promises to deliver in the coming months.
man i wish longhorn is out now so there would be winFS support for linux
with Windows file systems and this is fragmentation. For someone like myself who stresses XP quite heavily, the single greatest cause of system slowdown is fragmentation in excess of 10%, which seems to occur in 2-3 weeks.
get that ASUS board that has hardware CPU temp protection….an ATI card.probably what ever is in the 150s will be fine….a SATA drive, that will give WinFS some extra OMMF….a CD-RW drive and a DVD-RW drive (seperate because the DVD-RWs can get burned out faster)…front side firewire and USB2….and a nice 17 inch LCD screen, wide screen perhaps?
all enclosed in a nice looking case.
I tell ya, Windows will finally have the solid underpinnings to get some good work done.
WinFS is not a filesystem. It’s a service implemented on top of NTFS.
well, that is easily solved if they just take and apply some of the latest ideas in FS development….if they are aggressive with de-fragmentation automatically as part of the disk management system of windows, then when ever the disk is not doing work, de-fragmenting can take place…if it is done all the time, then the amount of time to defrag every time would be minimal.
the earlier hearsays about winFS is that it would replaces ntfs completely in the longhorn release. so by your comment are you telling me that longhorn still depends on NTFS and winfs is just a utility for NTFS?
I’d rather have OpenBFS for Linux. Unlike WinFS, it isn’t just kludgy patches on top of kludgy patches. it was built from the beginning to replace the propreitary BFS, which is about a bagillion times faster then NTFS, and offers quite a bit more features.
i wish there was more news about the HFS+ attributes Apple’s working on using, once again, another approach better then MS’s. its built into the file system, not a layer added to a layer.
all these WinFS is the best thing to happen to data since data article have got to stop. he says nothing has happened to file systems in the last 10 years (i assume he means journalling a little over 10 years ago). but BFS implemented a lot of the paradigms (no XML interface, because MS killed Be before XML was hot) he is touting as revolutionary but it was built into a single FS api, not a file system with a relational layer with a “core” layer, with a “service” layer, and finally an API layer. and unlike he states WinFS is not a file system, its a way to get at a file system. NTFS is the file system. without NTFS WinFS is nothing and that’s its terrible flaw.
What’s even funnier is OS X should have an attributes enabled Finder which also satisfies most of these paradigms well before longhorn is released.
instead of trying to catch up and reimpliment ideas that have already worked, MS should work on creating something truly innovative.
It’s in the article…
http://msdn.microsoft.com/library/en-us/dnwinfs/html/winfs03112004-…
 By Bannor99: Fragmentation on hd today don’t slow down machines as much as you think. I heard a interesting statement on todays fragmentation vs fragmentation 5 years ago. It was hd are fast enough that you would not notice unless you are doing high end video work. Defragmentation is pretty much as waste of time unless you work with super large files daily.
Here is a curious bit of trivia that I just found – apparently OS X.3 (Panther) has inbuilt auto defragging for files up to a certain size. In fact since this seems to be a kernel feature I assume that it would also be in Darwin.
“Everytime an application opens a file for reading, HFS+ checks if the file is fragmented and is less than 20MB in size. If so, it copies the file’s contents to a continuous region on the disk and frees up the previously allocated blocks”
If your interested to know where I heard about the defragment statement, it was on Your Mac Life, past show.
http://www.yourmaclife.com/
Looking at this, I see potentional for tremendous bugs and crashes in the well-known Microsoft style, as well as unecessary complexity and possibly even dangerous new security risks. The last thing we need is added complexity and bloat to an already bloated system.
I also wonder about the ominous capability of WinFS to recognize multimedia files, surely this is going to be tied into DRM and other extortion/policing schemes. Furthermore, clearing your usage histories will be near impossible if all your actions are being catalogued. This is not just a privacy problem it’s also a security threat.
The average person isn’t going to need or want WinFS, I sincerely hope there will be an option to keep it turned off.
WinFS sounds exciting and useful on paper, but I wonder how robust and compatible it will be in real life. It sounds like a very easy way for Microsoft to achieve even more lock in, which makes me nervous. I also don’t like the way it piggybacks onto the filesystem. I actually designed (but never fully coded) a system similar to WinFS, but abandoned it when I realized that functionality really should be built into the filesystem, like in BFS for example. I’m really hoping Apple comes up with a similar solution for OS X that really kicks ass.
Jared
http://www.theideabasket.com
instead of trying to catch up and reimpliment ideas that have already worked, MS should work on creating something truly innovative.
If you make something just to be innovative or “different” you’ll usually end up with something that’s barely usable or at least probably not better than the alternative.
You shouldn’t try to be innovative just for the sake of it. Besides Microsoft doesn’t have such a great track of innovation, they just bring innovation from the past to a larger crowd. That’s their thing.
That was transfer speed and sequential access speed which improved in HDD-s.
But random seek time is same as it was 10 years ago.
Nad file fragmentation is just thing which causes random seeking.
Huge RAM buffers in modern HDDs helps a bit, but not so much.
What about 2-3 weeks performance degradation complaints, i really suspect that author uses FAT32 under WinXP.
NTFS is quite resistant to fragmentation due its allocation algorythms. (Like BFS)
by bringing innovation from the past to a larger crowd, you mean bully the guy who came up with the idea until he can’t feed his family any more, steal his idea, and come up with a crappy implimentation of it, i agree completely.
by innovate, i meant come up with new useful things: actually study how people work most efficiently and naturally, how they think and relate to things, possibly reinvent the interfaces (they did an awesome job with the scroll wheel and affordable natural keyboards!)
on a positive note, i was going to add that if the relational layer caches indexes and metadata then you would get what appeared to be very fast searches and even normal file browsing… but then i remembered BFS did that already. man, WinFS is a joke.
While the overall idea of a database leveraged FS makes sense from the technical standpoint, I doubt this will really help anything. Most users I encounter can’t find thier files because they first have no idea where they saved it and next have no idea what it was named. Sure, they can search for things like “wedding pics, me and the wifey kissing”. Of course we have to assume that they first bothered to annotate the images to begin with. Will users do it? DOUBTFUL!!!
All we will have is a bloated layer on top of a relatively slow FS that no one but high end skilled users will ever benefit from. I can see the ad copy now. “Not only will Win Longhorn make your internet experience faster, better and more wholesome, it will let you forget about things like file names and directorys”.
End result? I will still have to show them that xxm.00000013.jpg in directory xmm-02-09-2004 is thier wedding photo that they uploaded from the Kodak Pic Disk!
WinFS will just make it more painful!
This is applying a high end fix to a low end problem. First show the user how to effectively use a hammer, then give them the nail gun!
All i want is a column to add a comment my files! To hell with fullblown databse system, just one silly property, is that too much too ask?
🙂
That, and a bit more, is available in Windows now.
Right-click on the file, go to Properties | Summary tab.
It is already possible enter comments for files and have them show up in the shell.
All files in NTFS can store additional data in seperate file streams within the filesystem.
You can enter the data by going to the files properties summary tab. There are fields for Title, Subject, Author, etc. and a large space for comments.
You can see the comments for each file as a column in details view in XPs explorer shell by going to View -> Show Details -> And checking the box for comments.
You can sort your files by their comments and even use the “Show in Groups” option to seperate your files into sections based on their comments.
I don’t think at the moment you can search by the comments field, though that can probably be done through the indexing service.
I would have been happy with metadata, the indexing service, and a more advanced search application. I agree with most of the negative comments I read here.
Quote:”I also wonder about the ominous capability of WinFS to recognize multimedia files, surely this is going to be tied into DRM and other extortion/policing schemes. Furthermore, clearing your usage histories will be near impossible if all your actions are being catalogued. This is not just a privacy problem it’s also a security threat.”
For me, it’ll be interesting how/if they implement this.
One major flaw is Windows’ reliance of file suffixes. If a file has an incorrect suffix (an avi file with a doc suffix for example)you could end up with Word having to load up a massive text file.
From what I’ve read, it will/can use Metadata, but what about older files that will not have this embedded within?
If the file has no suffix, how does it cope? Currently, windows has no mechanism to recognise files without suffixes. If an Amiga back in 1999 could do that then, why is it not possible with ‘The World’s Greatest Operating System’?