[Info-vax] Anyone interested in another public access system

[Bill Gunshannon]

In article <rdSdnTee6OfhV37UnZ2dnUVZ_jGdnZ2d at giganews.com>,
	drb at ihatespam.msu.edu (Dennis Boone) writes:
> > What would you cite as an example of a "modern" file system?  Most Unix
> > systems won't even commit your file to disk until they get around
> > to it.  One of the nice things about VMS is that what you write to
> > disk is actually committed to disk within a few seconds.
> 
> Not even close to true, at least for free unices.  Linux and *BSD
> have sync mounts which complete the write *before the call returns*.
> No "few seconds".  Sync mounting is optional, giving you the choice
> where you need it.  It's been a long time since I touched AIX, but
> I think jfs does it too.
> 
> > A power failure does not always require repairs to a VMS file system;
> > something that can't be said for Unix.
> 
> Again, not even close to true.  Various journaling file systems
> have existed for a long time - JFS in AIX, for example, ReiserFS and
> EXT3 in Linux, and others, which avoid the need for lengthy fsck at
> reboot time.
> 
> > Unix provides no means to create a contiguous file; just splatter it
> > all over the disk!.  Even <obligatory retching noises> Windows has
> > a utility to make your files and free space contiguous.  Unix just
> > doesn't care.
> 
> All over the disk?  Nonsense.  Unices have always tried to keep files
> (and their related inodes) contiguous or close together on the disk.
> Defrag utilities exist for some filesystems, and though coverage could
> be better, fixing it later with a defrag tool is a bass-ackwards fix.
> In any event, the idea that the system has to *force* the file to be
> contiguous can't be useful very often -- surely a best effort is better
> than an abort.
> 
> The idea that nothing new has happened in the unix filesystem world
> in decades is also a crock of bull; consider e.g. journaling (jfs)
> and log-structured (Reiser) filesystems, content-addressed storage
> (venti), networked systems like afs or coda, zfs, etc.  Most of the
> above *came out of* the unix world, rather than just being implemented
> there.  And I can't even begin to talk about Irix, HP-UX, or some of
> the cluster file systems.
> 

I was also going to post the rebutal you see above, but I have learned
from experience that it does no good here.  Tommorrow someone else who
knows nothing at all about Unix will post the same fud while complaining
that Unix people are always fudding VMS.

bill

-- 
Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolves
billg999 at cs.scranton.edu |  and a sheep voting on what's for dinner.
University of Scranton   |
Scranton, Pennsylvania   |         #include <std.disclaimer.h>