home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!europa.asd.contel.com!emory!wupost!gumby!destroyer!ncar!noao!arizona!arizona.edu!telcom.arizona.edu!leonard
- From: leonard@telcom.arizona.edu (Aaron Leonard)
- Newsgroups: comp.os.vms
- Subject: Re: cheap and not buggy tcp/ip for VMS
- Message-ID: <1992Nov13.144844.3983@arizona.edu>
- Date: 13 Nov 92 21:48:39 GMT
- References: <9211131301.AA21539@usppc.abb.com>
- Reply-To: Leonard@Arizona.EDU
- Distribution: world,local
- Organization: University of Arizona Telecommunications
- Lines: 25
- Nntp-Posting-Host: penny.telcom.arizona.edu
-
- In article <9211131301.AA21539@usppc.abb.com>, Dave@usppc.abb.com writes:
-
- [ excellent evaluation of TGV MultiNet omitted ... ]
- | The [TGV NFS] server also seems to take
- | longer to respond to directory querries than the SunOS NFS server on
- | machines of similar performance (SS 4/330). We are unsure of the cause.
-
- The problem is the different semantics between the native VMS and
- Unix filesystems. Unix expects the filesystem to maintain a precise
- byte count of each file; VMS Files-11 only knows block counts.
- Thus, when you do an `ls -l *' on a filesystem NFS-served by a VMS
- system, in the (frequently encountered) worst case, the VMS side must
- actually read all of the files in that directory in their entirety, in order
- to give a "proper" byte count to the Unix side.
-
- It would be nice if NFS clients were happy with block counts only, but
- I think that byte-count knowledge is ingrained pretty deeply in Unix.
-
- Aaron
-
- Aaron Leonard (AL104), <Leonard@Arizona.EDU>
- University of Arizona Network Operations, Tucson AZ 85721
- "It's not a bug, it's a form of flow control."
- - Jerry Leichter on why crash-prone Unix is a suitable
- platform for NSFNET core routers
-