[comp.unix.questions] Responses to NFS

ram@nucsrl.UUCP (Renu Raman) (05/15/87)

The following were the reponses(posted & emailed) about an earlier query
on NFS/Distributed File Systems(over long haul networks):

    1. Tait Cyrus posted about the work at U. of N.Mexico and LANL.
      (T1 carrier link between these two sites used for distributed simulations)

    2. Bill Sommerfield (MIT) elaborated on round-trip time on present
       generation links and packet size.

    3. Lindsay Marshall(I think from Newcastle) reminded me of the use 
       (or rather misuse) of Newcastle's registered slogan "UNIXes of the 
       world unite"**.  He had griped about NFS(*%@&#!@!~@%#$!@**).  
       I had used NFS only to identify it with distributed
       file systems.  Sorry, if that raised any hairs.

    4. Apollo is also using a T1 line, 61 miles across for a DFS.

    5. Joel McClung of Sun Microsystems mailed this to me. 

	 There is a product that Sun sells called Internetwork Router.  This is
	 some software that sits in a Sun and allows it to communicate over long
	 distances with another Sun running the same software.  This software
	 makes two LAN's appears as one logical LAN even though the physical
	 distance between the two LANs may actually be quite a distance apart.
	 If you want to communicate at speeds up to 19.2Kb then you can use one
	 of the serial ports on the back of the Sun.  If you want to go at
	 higher speeds, then you can put an SCP (Serial Communications
	 Processor) board in and go at a lot higher speeds.  I don't remember
	 the exact speed capabilites off the top of my head.  With this product,
	 you can NFS to your hearts content, although I believe you will have to
	 adjust the timeouts up a little to make a more robust connection.

    6.The following two were the most interesting:
      Michael J. North from Cornell informed me:

		Yes it has been done and is still going on.  Currently
	 we have a file system exported over NYSERNet.  RPI is exporting
	 it from a SUN-III.  We at Cornell have it mounted on our Gould 9050
	 and the University of Rochester has it mounted on either one of their
	 VAXes or a SUN-III (don't remember which).  

		The response is pretty good.  Hardly notice the difference
	 when you list the files.  We have all the RFC's being mounted
	 so there is one central location around here in which to get them.

		We've had it running for about 1 week.  No problems thus far.

		Mike

    7. The following predictions were made by some Berkely profs during
       a faculty retreat last spring(aren't we into summer now).  They
       were posted in "soc.culture.indian" sometime back.
       [I am not sure if this was a joke or a serious exercise in futurology]

	 Workstations -- Network
         ------------    -------
       a.By 1992 there will only two classes of computers: personal computers
         and supercomputers.  99% of us will use PCs.

              - Probable

       b.By 1992 the workstation market has crashed: no new applications can be
         found that would demand higher performance.

	      - Doubtful (when PCs are calling themselves as workstations,
	      wouldn't that mean this contradicts the first prediction).

       c.We must move beyond UNIX to integrated programming/writing environments
         that support multimedia data representations.

	      - Things are cropping up already.

    [--d.Remote resource sharing will become increasingly important  (running
    [    pieces of your AI(Yuk) program in Burma and pieces in Australia). 
    [--- [Guess, they will end up in pieces].

	------------------_____________________-------------------------
	Renu Raman				UUCP:...ihnp4!nucsrl!ram
	1410 Chicago Ave., #505			ARPA:ram@eecs.nwu.edu
	Evanston  IL  60201			AT&T:(312)-328-7145 
        __________________---------------------_________________________
you have mail

**This is a registered slogan of the Newcastle Connection 

P.S: The only prediction that I liked was:

     	"In five to ten years, any ``Survey Talk'' by one of our
	colleagues  at  a departmental colloquium will be understood
	by at most one other colleague and two graduate students."