This section describes some of the possible deployment methods you
can use for disk based clients, client controllers, simple
network and stand-alone systems. Unlike diskless clients, these systems
do not need to download images from a boot
server. However, before they can boot, they must be
installed with their operating system and application. There are
many ways to do this, but three of the
most common ways are:
LAN installation
EPROM blowing
Disk duplication
The deployment method you use depends on what external interfaces
your target machine has. For example, a disk based
client usually has a LAN connection and a floppy
disk drive, whereas a stand-alone system may not have either.
No matter which deployment method you use, you will always
have to install from an installation server (which may also be your build
machine). The software
to deploy must be installed on the installation server
first by the methods described in
``Server supplement installation''.
LAN installation
This deployment method is the most common and is used
in most client server environments such as POS, office
systems and so on. The basic requirements are that
the target system has a hard disk, a floppy
disk drive and a LAN connection to the installation
server.
The target system is booted using either a boot installation
disk generated on the installation server or by using
boot PROM.
The boot procedure connects the target system
to the installation server and then downloads the system
software. Once the software has been downloaded, the system
will be able to boot from its own internal
hard disk.
EPROM blowing
Using an EPROM to store the operating system and embedded
application is especially useful in systems where the application
is unlikely to change frequently. The type of systems
that benefit from this deployment method are systems such
as Network Computers and stand-alone systems. A single system
is installed using the LAN installation method described earlier.
Once installed, the image of the hard disk is
compressed and copied to the EPROM. The EPROM can
then be inserted in the target system. At boot
time the software image is decompressed and is copied
into an in-memory filesystem.
This type of deployment has been simplified by the introduction
of DHCP (Dynamic Host Configuration Protocol). This protocol enables
the system to have an IP address allocated at
runtime. This means that a unique IP address does
not have to be configured into the software image
before it is blown on the EPROM.
A variation on such a system is one where there
is also a hard disk. The operating system and
embedded application, for example Netscape™, remain on the EPROM
and the hard disk is used to store user
information. The net result of this is that the
root filesystem is effectively read only and the user
filesystem is read/write.
Disk duplication
Another deployment method that is used is disk duplication. This
method is most commonly used when there are many
target systems that share the same basic configuration.
Rather than use the LAN installation method
multiple times, only a single system is installed. Once this is done,
the entire image of the hard disk is duplicated
to a second hard disk of the same physical proportions.
Once the disk has been duplicated, it can be inserted in the target
system.
In the same way as for EPROM blowing, this technique has
been made easier by the introduction of DHCP.