ftp.jcu.edu.au/2014.06.ftp.jcu.edu.au.tar/ftp.jcu.edu.au/doc/ROUTEvsBRIDGE

Internet Message Format | 1993-10-17 | 9KB

From: Brad Cooper <[e-mail]>
Subject: What to do when you move PCs onto "wrong" subnet?
To: [e-mail]

At JCU I propose to install a 13-port router in the middle of our campus
fibre optic backbone and create 13 IP subnets. We are a small campus and
have survived very well without subnetting so far, but growth dictates
that ITS TIME! (that sounds vaguely familiar).

That leads me to crossing a bridge (router in this case) that I'm sure most
aarnet-contacts have already crossed and so I ask for advice that may
save me falling off the bridge (router).

How do you cater for the problem created by machines getting smaller and
more mobile (laptops) whereby a user at work is in one building on one
subnet with the appropriate subnet address and then he/she takes the PC to
a different building in a different subnet or perhaps dials in on a SLIP
line to the modem bank which is in yet another subnet?

Do others see this as a problem? If so, how do you deal with it? If not,
how do you deal with it? Also, if it is a problem, is the new Dynamic Host
Configuration Protocol (DHCP) designed to address this? I know DHCP is
designed to allow hosts to acquire their IP addresses from a server
database, but that is all I currently know about it.

Any help/advice/suggestions most welcome.

Regards, Brad Cooper        Internet:[e-mail]
Manager, Networking & Systems    Phone:[phone]
Computer Centre              Fax:[phone]
James Cook University
Townsville, QLD
Australia 4811


ANSWERS THAT AARNET CONTACTS SENT IN FOLLOW:

From: [e-mail]
==================================

BOOTP and RARP servers on the local networks may be of some help here, and
SLIP servers sometimes have dynamic IP number assignment capabilities.

Regards, Chris Chaundy (Technical Manager, Networks)

Information Technology Services, Thomas Cherry Building,
The University of Melbourne, Parkville, Victoria 3052 Australia
Phone:    [phone]                Fax:   [phone]
Internet: [e-mail]     (X.121 505233430003)

From: Peter Elford <[e-mail]>
======================================

This is usually handled using
dynamic address allocation from a server, typically using either
RARP or BOOTP. There ain't no other solution other than the various
mobile host proposals around (such as VIP).

There is a very good article on DHCP in the August 1993 issue of Connexions
(vol 7 no 8). All the draft documents are on munnari in the internet-drafts
directory:

draft-ietf-dhc-between-bootp-03.txt.Z
draft-ietf-dhc-bootp-01.txt.Z
draft-ietf-dhc-bootp-02.txt.Z
draft-ietf-dhc-options-03.txt.Z
draft-ietf-dhc-options-04.txt.Z
draft-ietf-dhc-protocol-06.ps.Z
draft-ietf-dhc-protocol-06.txt.Z
draft-ietf-dhc-protocol-07.txt.Z

Peter Elford
Cisco Australia, Canberra

From: Alan Agnew <[e-mail]>
=====================================

Brad,

Moving machines are a problem.

We just give these folks more than 1 IP address.

Fortunately there are not many of them, and we don't provide a dial-up
SLIP service.

I have heard that IETF are investigating the issue, I don't know
wheather DHCP is the answer or not.

Alan Agnew
QUT

From: Mark Prior <[e-mail]>
==========================================

We also use a single router on each campus as a hub. Although we
haven't got any mobile hosts yet what we will try to do is have a
machine running bootp on each subnet (that a mobile host may connect
to) and configure it to supply the necessary configuration info (IP
number, netmask, broadcast address, time servers, dns servers etc).

I think DHCP is a superset of the current bootp functionality but I'm
not sure what has been added.

Mark.

From: Greg Watson <[e-mail]>
==========================================

Peter writes...

> As you suggest in the next paragraph, this is usually handled using
> dynamic address allocation from a server, typically using either
> RARP or BOOTP. There ain't no other solution other than the various
> mobile host proposals around (such as VIP).
>
> > Do others see this as a problem? If so, how do you deal with it? If not,
> > how do you deal with it? Also, if it is a problem, is the new Dynamic Host
> > Configuration Protocol (DHCP) designed to address this? I know DHCP is
> > designed to allow hosts to acquire their IP addresses from a server
> > database, but that is all I currently know about it.
> >
> > Any help/advice/suggestions most welcome.

From personal experience, and in my opinion, any sort of dynamic IP address
allocation is a bad idea from a security viewpoint. Unless you are able
to identify a MAC address with an IP address and tie these both to a particular
machine, trying to prove (in court especially) that a particular individual
was responsible for hacking activity can be a very difficult task. You will
also run into problems if you have some sort of firewall protecting your
admin systems. How are you going to identify a particular machine that you
want to allow access to these systems?

Of course, none of this helps your original problem...

Just my $0.02 worth.

Greg

From: [e-mail]
=============================

Brad,

> At JCU I propose to install a 13-port router in the middle of our campus
> fibre optic backbone and create 13 IP subnets. We are a small campus and
> have survived very well without subnetting so far, but growth dictates
> that ITS TIME! (that sounds vaguely familiar).
    Why? What problem do you think it will solve (or prevent)?

> How do you cater for the problem created by machines getting smaller and
> more mobile (laptops) whereby a user at work is in one building on one
> subnet with the appropriate subnet address and then he/she takes the PC to
> a different building in a different subnet
    I cater for it on this campus (1300 nodes) by having a multiport
*bridge* in the middle of our campus network. After a careful study of the
pros and cons, I couldn't find any real need to use subnetting within a
campus, and it clearly causes problems such as you've outlined. Yes, I know
it's commonly done, but the reasons given for it are typically either
motherhood ("Everyone does it that way") or plain wrong ("it cuts down the
traffic", or "you'll drown in broadcast storms if you don't"). Not using
subnetting also cuts down on the effort required to support the network.

    This decision has been, er, controversial, but I haven't seen
anything yet to make me change my mind.

                                        Rollo Ross

Info Technology,  University of South Australia, The Levels, SA 5095, Australia
Ph [phone]    Fax [phone]   DTE 505282622004    [e-mail]

From: Simon Hackett <[e-mail]>
============================================

>         I cater for it on this campus (1300 nodes) by having a multiport
> *bridge* in the middle of our campus network. After a careful study of the
> pros and cons, I couldn't find any real need to use subnetting within a
> campus, and it clearly causes problems such as you've outlined. Yes, I know
> it's commonly done, but the reasons given for it are typically either
> motherhood ("Everyone does it that way") or plain wrong ("it cuts down the
> traffic", or "you'll drown in broadcast storms if you don't"). Not using
> subnetting also cuts down on the effort required to support the network.
>
>         This decision has been, er, controversial, but I haven't seen
> anything yet to make me change my mind.
>
>                                             Rollo Ross

Well, I've certainly seen enough things that have made me decide that
I prefer routing to bridging when the network starts to seriously
grow. I think it's the right way to go, and it does protect you from a
variety of problems that anyone on any of your cables can otherwise
inflict upon the entire campus. But, as they say in the classics,
"Your mileage may vary" :-)

Simon


{------------------------------------------------}
{  Simon Hackett,  Internode Systems Pty Ltd     }
{  E-mail: [e-mail]                }
{  Phone: [phone]  Fax: [phone]    }
{  Mail: PO Box 69, Daw Park, SA 5041 AUSTRALIA  }
{------------------------------------------------}

From: [e-mail]

Simon says (of routing versus bridging):

> I think it's the right way to go
    Motherhood! Motherhood!

> it does protect you from a variety of problems that anyone on any of your
> cables can otherwise inflict upon the entire campus.
> But, as they say in the classics, "Your mileage may vary" :-)
    Yes, it's a balance between those potential problems (which we
aren't experiencing) and the loss of flexibility, and the extra time
required to keep things working.

                                            Rollo Ross

Computer Centre,  University of South Australia, The Levels, SA 5095, Australia
Ph [phone]    Fax [phone]   DTE 505282622004    [e-mail]