Better file placement in Windows (vNext)

Tech-Archive recommends: Repair Windows Errors & Optimize Windows Performance

Hi,

I was reading this post on the Engineering Windows 7 blog: http://blogs.msdn.com/e7/archive/2009/01/25/disk-defragmentation-background-and-engineering-the-windows-7-improvements.aspx

And I was wondering why it is not possible to have Windows use the abondance of CPU power (compared to disk IO) to determine a more optimal file placement than the current, apparently very simple algorithm of: find free space, write file to free space, if not all of the file is written, repeat.

The idea, which probably isn't new or innovative, is explained on my blog here: http://www.larud.net/subtext/archive/2009/02/10/46.aspx

So I started looking at what it would take to do something like this, but this post already has put me off: http://www.tech-archive.net/Archive/Development/microsoft.public.win32.programmer.kernel/2007-01/msg00082.html

I'm a mere .NET LOB application architect, I have done my bit in DOS with Irq's and interrupts to talk to hardware, and I have come to understand that the NTFS file system driver is very complex, so that even attempting to prototype this is out of my reach. Maybe not intellectually, but I do have a day-job and a family ;-)

But Microsoft must have the engineers to be able to change it, right. So what I am looking for is some reasons why MS has not implemented a 'file placement strategy' like I describe on my blog. The target is purely desktop/laptop systems that people use daily and for many years. A file system for file- or application- or database-servers is surely going to work differently.

Maybe a newsgroup post is not directly the right place to ask, but as I have seen from the above mentionned post, some real knowledgable people hang around here who must have a better insight into 'why it is not done'.

Thanks for any feedback,
Rudi

.

Relevant Pages

  • Re: Persistence design [was: RE: OT: why are LAMP sites slow?]
    ... A blog piece of yours truly :-) ... There was a time when a database meant a flat file, ... Yes, the file system. ... metadata management means that the filesystem is now a better choice ...
    (comp.lang.python)
  • Re: File system and UML diagrams
    ... still i am updating) ... system (an OOA/D Model) as opposed to the data content of the files. ... File system. ... blog: http://pathfinderpeople.blogs.com/hslahman ...
    (comp.object)
  • Re: Swappiness
    ... Is there still something in FC6 to adjust swappiness like there used ... I couldn't find it in the /proc file system. ... Blog: http://thecodergeek.com/blog/ ...
    (Fedora)