On Thu, 2 Mar 2000, Bjorn Wesen wrote:
> On Thu, 2 Mar 2000, Richard Torkar wrote:
> > Well, actually AFAIKT, it's not symbolic links as we have in UNIX-land
> > that's the great part.
> > As I understand it it's the server part actively searching for identical
> > files and then symlinking them, i.e. automagically.
> >
> > What I can't figure out is, why it took them so many years, and are they
> > really first?
>
> I don't know their implementation, but it sounds like 'copy on write' but
> on disk instead of in RAM - and with a tool that merges the blocks or
> files into a COW-marked block/file at regular intervals.
I'm reminded of the literal pool in MACRO-10. Whenever you go to store
some bits, look to see if they're already there, and if so then just point
to that copy instead.
> The big question is of course, why they HAVE so much redundancy on disk in
> the first place in their "flagship OS", that they need to resort to things
> like this to avoid it :) I mean, if redundancy really is that big a
> problem on a Linux box, we would have had compression techniques like this
> long ago (which it's all about - it's compression, but disguised in
> abstract files and "stores" instead of bit-string matching).
It must be because at least 90% of kitting teams for Windows products
don't know where to stash the .DLLs, so you wind up with 69 separate
copies of the exact same .DLL scattered all over. I can't think of any
other place where you'd find lots of duplicated files that hang around
long enough to be worth the bother.
Anyway the excitement isn't so much about the links as it is about what
they link to and how it got there.
-- Mark H. Wood, Lead System Programmer mwood@IUPUI.Edu "Where's the kaboom? There was supposed to be an Earth-shattering kaboom!" -- Marvin Martian, 01/01/2000 00:00:00- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Mar 07 2000 - 21:00:15 EST