Cad Guidebook: A Basic Manual for Understanding and Improving Computer-Aided Design (9 page)

Table 2.3 presents a simple comparison between some common storage
system’s characteristics. Some of this information will be explained in more de-
tail in Section 3.2: The Operating System.

Computer Hardware Basics 23

TABLE
2.3

A Simple Comparison of Various Computer System’s Storage

System type Disk drive designation Example file names

PC A: (typically floppy disk)

C: (main hard drive)

F: through Z: (typically

network drives)

DRAWING.DAT
Model_Of_A_Car.PRT

Unix™ workstation /dev/hdisk0 drawing.dat

model_of_a_car.prt

VAX US E R1:

USER2:

DRAWING.DAT;1
MODEL_OF_A_CAR.PRT;32

Mainframe A (user’s read–write area)

H (shared “virtual” disks)

(also referred to as DASD)

DRAWING DATA A1
MODLCAR PART H/A

2.6 NETWORKING

So far, the basic internal workings of a “standard” computer system has been pre-
sented. The essential elements of the system are the CPU, Memory, and Storage
systems. However, there needs to be means by which the system is actually put
into use and manipulated by programmers and users. This is accomplished by
devices known as peripherals. The familiar devices that provide interactivity with
the system are keyboards, monitors, pointing devices, speakers, etc. These types
of devices will be discussed later in this chapter. One important peripheral for the
basic functioning of a computer system is the Network Interface Card (NIC) or
network card. This circuit board provides a connection between the computer
system and a wider environment of computers and devices. This wider environ-
ment is called a computer network or just a network.

Over the years, the relative importance of the computer network has
steadily grown. At one time, a CAD-based computer would have been seen as a
self-sufficient unit that occasionally communicated with other computers. How-
ever, by the 1990s the network was seen as an essential element of any computer
system, particularly since the mainframe ceased to be the foundation of a compu-
tational framework. Instead, the foundation shifted to many smaller PCs and/or
workstations. As that happened, the network was seen as the appropriate means
to share resources. At first, this would be just for peripherals such as printers (an
office with a dozen PCs could share one expensive printer). Then, the systems
started to share data and storage. This meant that just one copy of a PC program
could be loaded on a designated central computer system (instead of having to
load the PC program to a dozen separate PCs).

24 Chapter 2

Although this new “file server” approach to networking sort of looks like
the mainframe system again, there is a significant difference. With the mainframe
computer, the data was indeed stored in a central storage system (similar to the
“file server” of network), but all CPU power also had to come from that same
mainframe computer. With the standard network approach, only the files came
from the central storage, while the actual execution of the programs was still
done “locally” (i.e. using the CPU in the user’s computer). The network approach
means that the computational power demands of different users do not affect each
other. So, if one user is running a CAD program that fully uses his CPU, it has no
effect on other computer (CPUs) on the network that may not be running CAD
programs. This would not be the case for the mainframe where all users are
“competing” for the centralized CPU capabilities.

2.6.1 Local Area Network (LAN)

The kind of computer network that uses the central computer as the file server
and/or the printer server is loosely referred to as a Local Area Network (LAN).
The LAN was primarily a local network since the devices that connected to the
network had very short distance requirements (such as no 2 computers could be
more than 110 yards apart). However, over time, these restrictions became less
and less severe, and LANs could grow to cover entire company sites and have
1000s or more computers and devices connected to the network. So, in general,
most company-wide networks can just be considered LANs, and most commer-
cial CAD systems in use are dependent on this type of network.

2.6.2 Wide Area Network (WAN)

Another class of network for connecting very wide spread sites (say at various
locations in a country) is the Wide Area Network (WAN). At one time, this net-
work would have used a entirely different type of hardware and software, but in
many cases, the LAN type of approach can be adapted to the very long distances,
but just with a vastly reduced performance. For instance, if phone lines are used to
connect the different office sites, the speed at which the network data can be trans-
mitted from one computer to another would be many times slower, but the user
may see no difference in how the computer file is stored (it is “transparent” to the
user). Since these WANs are generally so slow and CAD programs are generally
so demanding on computer system performance, there would be few CAD sys-
tems that are really dependent on this type of network (at least for interactive use).

2.6.3 The Internet

Finally, the “network of networks” has arisen called the Internet. It would be easy
to think that the advent of this worldwide network was a natural progression from

Computer Hardware Basics 25

the WANs. But most, if not all, of the successful LAN networks were based on a
system called Ethernet™. The Ethernet system was based on a networking proto-
col called IP or the Internet Protocol or TCP/IP. The way that all the Ethernet-
based computers communicated is generally the way that computers on the Inter-
net communicate. Although other types of communication can be achieved with
the Internet, virtually all of it is done in the Ethernet style, and therefore, virtually
all of the computers on networks communicate via an Ethernet card.

Although the speed of Internet communications can be very limited in
comparison to a LAN, and thus the use of the Internet for CAD programs should
be ignored just as it would be for a WAN, the advent of the World Wide Web put
an interactive and graphical face on the internet. The addition of this graphical
interface is made possible by various techniques that optimize the transmission of
graphical types of data (despite the slow network speeds). Since CAD programs
are very interactive and graphical, these optimization techniques can be applied
to the running of a CAD program “across” the Internet (although the user’s local
CPU again is intended to provide most of the computational resources). Although
this may not be the standard approach, it is likely to become more common as the
Internet continues to improve.

2.6.4 Network Functioning

As mentioned earlier, the device that connects a computer system to a network is
called the network card. This device may actually be a circuit board for the sys-
tem, or it may be integrated into the main board or “motherboard.” In any case,
when the computer system needs to communicate with the network (to get a file,
a program, or a shared device such as a printer), it will do so with the network
card. As with other devices attached to the computer, there are specific locations
in the memory map designated for the network data interface. Thus, the com-
puter’s software (such as a CAD program) usually does not treat the data from the
network any differently than data from a disk drive. However, the way in which
the network card actually acquires the data from the network is quite different,
and how it is done can greatly impact CAD program performance.

Devices such as disk drives are pretty tightly “coupled” to the computer
system in which it resides. The exact size and capability of the disk drive is
tracked by the system, and when the software running on the computer creates or
accesses the data on the disk drive there is no “competition” for that data. The
disk drive is totally dedicated to the operation of that particular computer. This is
not the case with the network. The nature of the network is that there can be any
number of computers and devices connected to it. A network is intended to be
dynamic. Therefore, the computer system that needs to get data to or from the
network does not really know what is happening at any given moment, and the
resources on the network can not be considered to be dedicated to the computer

26 Chapter 2

that wants it. So, the network card needs to be much more involved in the process
of retrieving the data. It must communicate with the network with its own pro-
gramming (usually part of unchangeable software stored on special computer
chips called ROM chips), and coordinate the translation of the data to or from the
computer system.

The way that the Ethernet-style networking system accommodates the dy-
namic behavior of the network is through a kind of broadcasting. One can think
of the network cards as “listening” to the network cable that it is attached to. It
listens for information that belongs to it. Or, the network card may become the
“transmitter” sending out data for other computers or devices. When the signals
are sent out to the network, the originating computer does not really know if the
signal is going to make it or not. If one computer sends out its signal at the same
time as another computer, they may collide. Each of these competing computers
then waits a short interval (with some intentionally random pause) and then tries
again. The message will eventually get to its destination (which is then acknowl-
edged by the computers communicating). It may seem that the collisions will be
prohibitive to reasonable performance, but the actual chunks or packets of data
that are attempted for each transmission are rather small (with the network card
appropriately cutting up the data that the computer wants transmitted), and if the
network is working fast enough, the collisions can be of no serious consequence.
This “collision-retry” type of network can operate at 10 million (10 Base T), 100
million (100 Base T), 1 billion (or gigabit) bits per second.

Although the network card and network cabling are essentially all that a
user needs to know about network functioning, there are many other “behind the
scenes” elements to a typical network. Some of these elements would be hubs,
routers, bridges, gateways, repeaters, and multiport adapters. Each of these de-
vices are connected to the network and each has a specific task to perform in get-
ting the data from the transmitting computer to the receiving computer or device.
Particularly with something like the Internet with its countless computers or
nodes, one can imagine there is a great need for shuttling, shuffling, and redirect-
ing of the network data to get it to the right place as efficiently as possible.

2.6.5 Network Performance

There are a number of very important ramifications of networking on programs
such as CAD systems. The first issue is network traffic. Considering the way that
data is transmitted with the “collision-retry” system, one can see that the perfor-
mance of the network (in terms of its ability to transmit those large CAD files
from a file server to a user’s workstation) is going to be greatly affected by how
many users are attempting to do the same function. If many users are sending/re-
ceiving at the same time, then the collisions are going to be much more frequent,

Computer Hardware Basics 27

and users will notice that operations take longer. Of course, the operations that
take longer will generally only occur when the user is saving a file. Recall that
the network approach is to use the user’s local CPU as much as possible to do
calculations and graphical operations (like displaying a drawing or 3-D model),
so network traffic should only be affected when the user is calling up or retriev-
ing or loading a drawing or model, and then when the user is saving a file.