Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site wdl1.UUCP
Path: utzoo!linus!decvax!tektronix!hplabs!hpda!fortune!wdl1!jbn
From: jbn@wdl1.UUCP
Newsgroups: net.unix-wizards
Subject: Re: Re: SUN ND Implementation?
Message-ID: <630@wdl1.UUCP>
Date: Thu, 8-Aug-85 21:26:59 EDT
Article-I.D.: wdl1.630
Posted: Thu Aug  8 21:26:59 1985
Date-Received: Mon, 12-Aug-85 06:18:30 EDT
Sender: notes@wdl1.UUCP
Organization: Ford Aerospace, Western Development Laboratories
Lines: 18
Nf-ID: #R:sun:-255200:wdl1:64000006:000:891
Nf-From: wdl1!jbn    Aug  8 13:13:00 1985


     Mt. Xinu is talking about supporting NFS on the VAX.  But they aren't
shipping product yet.
     Actually, the idea of a big machine as a NFS back end is probably a
bad one; NFS is compute-intensive.  We have ten diskless stations on a
SUN file server, and seem to be running out of cycles on the file server
long before becoming I/O bound there.
     What really seems to be needed is a port of NFS to an intelligent
disk controller.  The file server doesn't need to run UNIX; it doesn't
even need to run user programs.  The ideal would be a card like the SUN
2/50 card but with an SMD interface, no display interface, CPU(s), RAM, ROM,
no paging, and an Ethernet controller, with NFS in ROM.  You would have one
of these for each disk spindle, and each would have enough power to run
NFS for one disk spindle without becoming compute bound.  This might make
NFS work.

				John Nagle