Cad Guidebook: A Basic Manual for Understanding and Improving Computer-Aided Design (10 page)

Obviously, if many users are using the same network at the same time (so
simultaneous filing is not really avoidable), then the situation can be improved by
increasing the speed of the network. Another improvement might be to reorga-
nize the network so that each user’s computer has its own connection to a higher
speed “backbone” (as opposed to having groups of computers connected to inter-
mediate devices, and then the intermediate devices accessing the file server).

Another consideration for performance of the network is the way the file
server is utilized. Consider that even if the network is fast enough to not experi-
ence the “collision” situation, but all the drawings and models for CAD users are
stored on one disk drive on the file server, then all users must wait for that disk
drive “one at a time” regardless of the network’s capability. Another important
consideration for the network file server is the amount of memory or RAM in the
file server. This memory is used to hold or buffer data that is coming from the
network for reading and writing (like the cache mentioned earlier for storage sys-
tems), so as much RAM as possible should be used. The file server can become a
“bottleneck” in that it is the slowest or most constricted part of the network archi-
tecture. Clearly it is very important that an appropriate level of expertise be
brought to bear on the configuration of the network architecture.

2.6.6 Network Addressing

The way that the network recognizes which information is intended for a specific
computer (or “node”) is by an address. This is not the same kind of address that is
associated with the memory system of a computer. The address system for the
network starts with the network card itself. Each network card is given a unique
number. This is usually called the Ethernet address. It would be made up of 6
bytes of data (such as 08:00:11:00:AF:08). There is a system to the bytes (a byte
being 8 bits) assigned, but that is beyond the scope of this work. The program-
ming that runs on the network card’s components uses this number sequence to
communicate with the network. This address can be considered part of the “phys-
ical layer” of the network architecture.

With the physical address for the network card identified, it can then be
translated for use on a specifically designed network (such as a LAN or the Inter-
net). That network card could be placed in different computer systems over time,
so there is a higher “level” for assigning the address to associate with a computer
system. This higher level address is called the Internet address. This address im-

28 Chapter 2

plements a sort of network layer, and it uses the Internet Protocol (IP). The net-
work layer then communicates with a transport layer, where the Transport
Control Protocol (TCP) takes over. The way these layers work is beyond the
scope of this work, but terms like TCP/IP are in rather common usage, and it ba-
sically means that this Ethernet-style of networking is running the network. In the
end, a layer communicates with the actual application (such as the CAD system).

The Internet address has the form of 4 bytes of data. It would be presented
in a form such as the following:

128.1.128.157

This address uniquely defines a computer or device on the network in terms
that are understood by the software that really controls the network traffic. The
devices that “run” the network (such as hubs, routers, servers), or at least the soft-
ware running on these devices, use the numbers in the address to determine how
to sort, redirect, transmit, etc. the “packet” traffic. These devices are not con-
cerned with the data in the packets, just where it needs to go.

There are 3 “classes” to these addresses depending on the kind of network
architecture. These classes determine how many networks could be found in the
entire site, and then how many specific computers or devices could be found in a
given network (in other words the class defines to what degrees a network of net-
works can be created). The Class A address uses 7 bits to define the network and
24 bits to define specific devices. So, for Class A, there could be 128 networks (2
raised to the 7th power and including zero as an address), and there could be
16,777,216 specific devices on a given network. The remaining bit indicates the
class itself. At the other end of the spectrum, a Class C address allows 2,097,152
networks, but only 256 specific devices. Clearly, the addressing scheme used needs
to properly fit within the physical network architecture. If the addresses are used to
create “sub-nets,” then the proper network electronic devices need to be used to
keep communications within a specific subnet optimized. For instance, there may
be 3 different groups of CAD users at a site, and they do not generally need to
communicate data beyond their small group. In this case, the use of special ad-
dressing schemes can be an advantage. However, if all the CAD users need to com-
municate with each other pretty equally, there may be no benefit to this approach.

At some point, the Internet address is translated to a node name or host
name. This is a “real” name in the sense that it does not have to be in a numerical
format, and users would be more likely to remember it (for instance, serv1, cad-
meister, zulu, etc.). The host names are usually kept in a file that resides on each
computer system (for instance a “hosts” file).

2.6.7 Peer-to-Peer Architecture

An interesting effect of the Ethernet-style of networking is that it is usually con-
sidered a “peer-to-peer” network architecture. This means that all the computers

Computer Hardware Basics 29

on the network are considered “equals” (at least if there were no security schemes
applied). Therefore, a user can access other computers on the network to some
degree. This is done with programs such as rlogin and telnet. Using rlogin (mean-
ing remote login), one could have a program that does certain calculations run on a
different computer. In this case, the user’s originating computer will not slow down
due to the calculations consuming CPU or storage system resources on that remote
system. Obviously, this can cause security problems, but it is a useful capability in
a CAD environment where some “batch” processing may be needed for computer
programs that run for hours or days. Some other programs that are typically uti-
lized across the Ethernet-type network architectures are shown in Table 2.4.

2.6.8 Network File System

The final subject to be addressed with respect to networking and its importance to
the computer hardware configuration for CAD software is the network file sys-
tem (or NFS®
or NTFS). An NFS-type program is basically a “trick” to make
network files look like they are local files. For instance, a workstation may have
one disk drive attached to it that contains most of the software that runs on that
computer (such as the CAD program). When the CAD program runs (using the
local CPU resources, of course), it may look for files that are identified by a spe-
cific directory and file name. This name may actually be pointing to a file on a
different computer on the network (often to a computer called a file server). It is
this NFS which makes the network-based files look like they are local files. Set-

TABLE
2.4

Common “Peer-to-Peer” Network Utility Programs

Program or

command Capability

export DISPLAY Allows graphical user interface to be sent to another workstation as

if it were seen on the user’s current computer.

ftp File transfer protocol (ftp) is an Internet based program to transmit

files across a network. If a network node is to store files for oth-

ers to download, then it may be referred to as an ftp site.
ping An Internet based program to see if a remote computer is respond-

ing to the network.

rcp A program to interactively copy a file across a network following

the typical copy command for a Unix workstation.

rlogin A program to mimic being a local user of a remote computer on the

network.

rsh A program to initiate a system command on a remote computer fol-

lowing the typical process for a Unix workstation.

telnet A program for gaining access to a remote computer on the network.

30 Chapter 2

ting up the mapping of how local names will actually be redirected to network-
based names can be called mounting directories, making NFS mounts or map-
ping drives (depending on the type of computer system).

The NFS technique can be quite useful in a CAD-based network, but it
must be used carefully. Recall that in the earlier discussion of the memory sys-
tem, there was something called “virtual memory” that allowed a disk drive to
make the computer system have more memory capacity than was really available
on the memory chips. The part of the disk drive used for this purpose must be as
fast as possible, and therefore it should be made available on a local disk drive.
Using an NFS type of access for this “swap file” would not work very well. An-
other poor use of NFS would probably be the operating system files. These are
files that the CPU constantly needs, so putting it “across” the network would
probably cause problems as well. In most cases, the CAD software would also
generally be a poor choice to put on the file server. While the CAD software is
running, the CPU may continually need to retrieve information from the disk
drive for the software; therefore, a “local” drive would probably be best (unless
the central management of the software was an overriding concern).

A good choice for the NFS technique would be data that is only needed
occasionally or data which needs to be centrally stored, managed, and protected.
The actual drawings or 3-D data files that a CAD program creates are a good
candidate for storing on the file server. In this case, the file is really only accessed
intermittently (when the user first loads the drawing or model and then whenever

FIGURE
2.5

Typical CAD system network arrangement.

Computer Hardware Basics 31

the user saves it again). If these files are all stored centrally (on the file server),
then it is relatively easy to perform system management operations (such as
“backups”). Of course, users need to realize that although the files appear to be
on a local disk, their use is actually going to generate network “traffic” that af-
fects other users. Figure 2.5 shows a typical arrangement of data and files for a
CAD system network.

2.7 PERIPHERALS

This section presents information on the devices that are attached to the com-
puter. These devices are the peripherals: the monitor, keyboard, pointing devices,
and hardcopy devices.

2.7.1 Monitors

Probably the most important peripheral for computer systems that are used for
CAD software is the monitor and its counterpart, the graphics adapter. The very
essence of CAD software is the use of computer graphics. The designer or engi-
neer interacts, calculates, and specifies in graphical terms. Therefore, it follows
that the device that creates these images is extremely important.

The monitor is simply the device that shows the images from the software
(see Figure 2.1). However, it is not the device that creates the image; this is the
graphics adapter discussed in the section below. All that needs to be understood
at this point is that the graphics adapter sends the proper electrical signals to the
monitor so that it can show the correct image.

The basic characteristics of the monitor are its physical size, resolution,
and ability to react to the electrical signals from the graphics adapter. The physi-
cal size is generally based on a “diagonal” distance of the viewable area of the
screen. This diagonal distance, for instance, would be from the top left corner of
the screen to the lower right corner. Obviously, the larger this distance or di-
mension, the easier it is to view the image on the screen (assuming the graphics
adapter keeps the same precision of the image) since the image is larger. For
CAD software, the screen should be as large as is affordable (for example, at
least 500 mm or 19 inches for most users, or even larger for those with vision
disabilities).