1.1 THE CONCEPT
The idea behind running a diskless workstations is to run a complete operating
system with logins
and applications etc. without the need of purchasing another hard disk
for said workstation.
There are trade offs however. There is some disk space used on a server
The files on the server are network mountable therefore the network
gets extra load and
because the diskless workstation has no disk there can be no swap memory.
With these trade offs accounted for there is still some scope
to play with.
The possibilities are many.....
A small system for learning from or experimenting with.
A standalone system that operates only one function that does not warrant
the purchase of a
disk. eg. masquerade firewall, X-terminal, print server, etc.
This talk will aim more toward the X-terminal feature but hopefully
will cover any other
feature that you may require. Of course the instructions that
come with the software go into more depth than this discussion so refer
to them if confused or if you wish to know more.
1.2 BOOTP AND TFTP THE DISK SAVIOR
Bootp is the name given to the internet bootstrap protocol. The Idea is
that a main server, with
all of the operating system and boot strap code, listens to the network
for a request from a diskless workstation. The request being "Please boot
me." (layman's terms :-)
If the main server is satisfied that the requesting machine is one
that it is allowed to boot then it sends bootstrap code back over the network,
using tftp, to the requesting machine.
Tftp stands for Trivial File Transfer Protocol and is similar to the
normal file transfer protocol but does not contain password
protection when a system asks for a file.
This lack of security is of concern on a server so the server can be
set up to allow file transfers on only one directory tree.
The default directory at the time of writing was /tftpboot/.
1.3 NETBOOT (for Linux)
Netboot is the name of the code developed by Gero
Kuhlmann for setting up the Linux system to remotely boot a diskless workstation.
There are three major steps in developing the system.
1. Create code for a floppy disk or EPROM that is placed into a diskless
This is the code that requests for bootstrap code from the main server
over the network.
2. Create bootstrap code that can be sent over the network. This
is the full kernel with all
the bells and whistles.
3. Create the file system on the main server that the diskless workstation
(Yes some disk space does get used but it is still a lot more economical
complete new disk).