NetNews Usenet Archive 1992 #30/NN_1992_30.iso/spool/comp/unix/admin/6646

Text File | 1992-12-15 | 2.6 KB | 56 lines

Newsgroups: comp.unix.admin
Path: sparky!uunet!spool.mu.edu!yale.edu!jvnc.net!tiger!kagumac2.cc.kagu.sut.ac.jp!user
From: [e-mail] (Devendra Narayan)
Subject: Planning for a 10,000 user network !
Message-ID: <[e-mail]>
Followup-To: comp.unix.admin, comp.sys.sun.admin
Sender: [e-mail]
Nntp-Posting-Host: kagumac2.cc.kagu.sut.ac.jp
Organization: Science University of Tokyo
Date: Tue, 15 Dec 1992 08:24:07 GMT
Lines: 43

    Our university is now planning for a large network of Unix
workstations. There a few questions I have about large scale
networks and experiences that other large sites have.

    We plan to have about 700-1000 workstations in our Tokyo campus
with 7,000-10,000 students having accounts on these machines. From
the administrators standpoint, what we would like to have in the
system is that any of these users should be able to log onto any
of the workstations and access their unique home directory. We would
have to use NIS ( ? ) and NFS. Whether we have all the information
on one large ( Super ? ) server or a couple of smaller servers will,
of course, depend on the cost - performance of the server(s).

    I am seeking information about the comparative performance of
various hardware manufacturers server machines as well as on
issues that we will have to be concerned with - network speed
and performance ( whether to use multiple ethers or FDDI or 100
baseT or HIPPI or whatever ), the server CPU and internal bus
performance, whether to use a single, large mainframe or even
super-computer level machine or perhaps distribute the whole
information ( all the 10,000 homes plus the NIS info. ) over
several servers. From the user's standpoint, all this has to be
transparent so that when he/she logs in to any workstation, he
sees the same home directory. There are also issues of security
etc. that probably need to be considered. NIS apparently has an
upper limit of 2500 users and I would welcome any information
on NIS replacements.

    I seek input from unix administrators from large academic sites
who probably have already faced ( and hopefully solved ) some if
not all of these issues. Would really appreciate any and all input.
I do not know if there would be enough interest in the group about
such matters, but if there is, perhaps replies to the newsgroup
will generate more interest and be useful to others as well.

Thanks
Devendra Narayan

===  = = ===       Devendra Narayan
=    = =  =        Education Division,
===  = =  =        Information Processing Center
  =  = =  =        Science University of Tokyo
===  ===  =        E-mail : [e-mail] or [e-mail]