On Feb 18, 2008 4:17 PM, Mark Martin <[EMAIL PROTECTED]> wrote:
> On Feb 7, 2008 12:50 AM, Darren Reed <[EMAIL PROTECTED]> wrote:
> > If that's something you'd like to pursue yourself, feel free :)
> >
> > But I'm curious to know, have you looked at the "locate" program
> > from GNU's findutils?
> > http://www.gnu.org/software/findutils/
> > locate and updatedb are the tools of importance here.
> >
> > Without thinking about what others do, what is it that you
> > want to do that you can't today?
>
> Treat the filing system as a set of structured documents, with indexed
> (and easily queried) meta-level information.  I'd like flexibility to
> extend types, and present "views" on my storage according to arbitrary
> hierarchies and order.

So, what sort of storage systems would you like to use
this on? Desktop or server? Personal or shared? Single
system or enterprise network?

For example, higher levels of abstraction would be really handy on one of
my thumpers. I'm looking at 100 million files. For example, with images
I need to know the file type, color depth, resolution, and (for us) the
copyright statement contained in the image. It would be nice to know
the relationship between images, and probably some sort of checksum.

I guess this would be 100 bytes per file (the raw data is more, I'm assuming it
compresses to 100 bytes). So that's about 10G of raw data. Index it for
searching and it's rather more. Do that for all our servers and it starts to
become a non-trivial problem (which is why I haven't just shoved the lot into
a mysql database and had done with it).

Still, a structured database you could query would be hugely better than a
script that runs 'identify' on 100 million images to pick out the ones that
have certain attributes :-)

-- 
-Peter Tribble
http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss

Reply via email to