Windows and TCP/IP for Internet Access - Dersindir

gazecummingNetworking and Communications

Oct 26, 2013 (3 years and 9 months ago)


Windows and TCP/IP for Internet Access


nternet, the world
wide network of computer networks, has captured the imagination of
the general public. In mid
1993 the Internet was barely mentioned in the popular
computing magazines. By mid
1994 it
was the topic of articles in national news
magazines, local newspapers, and grocery
store tabloids. By Autumn 1996 the Internet
was the phenomenon of the decade.

wareness of the Internet was spread initially by word of mouth. Computer pundits were
not dis
cussing the Internet in Spring 1993 when I first began investigating the Internet in
my work as a librarian. Indeed, most pundits seem to have acquired Internet access only
in the Spring of 1994. Prior to that time, computer magazines were not helpful for
wishing to learn about the Internet.

y December 1994, there was something of a feeding frenzy of interest in the Internet.
Bookstores were flooded with guides to the Internet. Software vendors were rushing to
market with collections of software des
igned for navigating the resources on the Internet.
It was almost as if the crest of the Internet wave had passed. Pundits who did not have
access to the Internet in 1993 were in late 1994 already writing negative opinions about
the difficulties of navigat
ing Internet resources, and about the uselessness of those
resources. This frenzy of media interest, praise, and criticism continues to this day.

omplaints about the Internet are many. Certainly it can be difficult to find information
and resources on th
e Internet. A great deal of information is unvalidated, non
authoritative, or otherwise questionable. Some is deliberately false. Some resources
should not be available to children. Some would argue that even adults should not have
access to some Internet
resources. In other words, Internet resources are similar to all
other information resources in the world.

Internet analagous to streets of the city

t is important to remember that the Internet is not a service. Rather, it is a means of
gaining access to
services and of retrieving information and other objects that can be
represented electronically. In trying to understand the Internet, one might draw an
analogy between the Internet and New York City.

ew York is big, complicated, and disorganized. The ci
ty's myriad resources can be hard
to find. Some of what happens or what is available in New York should not be seen by
children. For those wishing to navigate the complexity of New York, there are
guidebooks, phone directories, magazine articles, and indiv
iduals with expert knowledge
about areas of particular interest. One can navigate the complexity of the city by subway,
taxi, and bus. One can even hire a private guide to conduct a tour of the city.

he Internet can be compared to the streets of New York

City. The services available on
the Internet have their analogies in the city's libraries, department stores, bookshops, art
galleries, street vendors, and street
corner zealots passing out literature or lecturing the
passing crowds. It is safe to assume
that somewhere on the streets of the city there will be
found information and services of interest to almost anyone. However, finding that
information might take some time for someone who is new to the city and its resources.
Similarly, somewhere on the In
ternet there also will be found information and services of
interest to almost anyone.

raveling on the Internet requires only a few basic tools. First is a computer with a
network connection to the Internet. A direct connection to the Internet uses a PC
equipped with a network interface card that connects to a local area network that is in
turn linked to the Internet. Such a connection is common at universities, and becoming
common in businesses.

f a direct network connection is not available, an altern
ative is to connect to the Internet
through the computer's serial port. This involves a telephone connection to a terminal
server that offers SLIP (Serial Line Internet Protocol) or PPP (Point to Point Protocol)
service. Any of these connections can be use
d with a variety of commercial or shareware
software to make your local computer a host on the Internet and to access services and
information from the entire earth. This tutorial emphasizes the use of freeware and
shareware versions of software running un
der Microsoft Windows 3.x and Windows 95.

Windows and TCP/IP for Internet Access


s recently as 1993, Internet services were accessible primarily through character
interfaces using a variety of complex command sets. Best
selling bo
oks on the Internet
contained page after page of screen displays or command sequences captured from
based systems executing basic Internet functions. By 1997, the World Wide Web
came to dominate the popular view of the Internet as being a multimedia e

efore the Web, the Internet services of interest to most people consisted of four basic
functions. These were electronic mail (e
mail), Internet news, file transfer between
computers (FTP), and remote login to another computer (telnet). Gophe
r was developed
as an access system to make it easier to browse the Internet and to search for relevant
information in a user
friendly manner. The invention of the World Wide Web quickly
displaced Gopher as an interface to Internet services. Yet e
mail, FT
P, news, and telnet
remain important functions for most people.

ffordable Internet software for Windows first became available in Spring 1993. Prior to
that time, Windows users were dependent for Internet access on expensive, proprietary,
commercial produ
cts in which each vendor's offerings were mutually incompatible with
every other vendor's offerings. Publication of the Winsock applications programming
interface (API) provided a way for individual client software (such as a telnet or FTP
client) to be co
mpatible with every vendor's networking products. As a result, beginning
in 1993 there was a blossoming of freeware, shareware, and commercial Internet software
for Windows.


he World Wide Web was developed by the high energy physics commun
ity to
distribute technical papers and other forms of data. It is a system that enables users to
find and retrieve information by navigating a system of hypertext documents. In a
hypertext document, selecting a highlighted word or phrase causes a new docum
ent to be
retrieved and displayed. Educators, businesses, and hobbyists now use the Web to
distribute multimedia information to a world
wide audience. Those interested in using the
Web for teaching or publishing may find it useful to read my paper "Teachin
g and
Publishing in the World Wide Web." It is available in both a
plain text version

and a
WWW hypertext version

he World Wide

Web is the ultimate toy for those addicted to changing what they are
looking at rather than viewing what they are looking at. It is the next step in addiction for
those who have grown bored with using their TV's remote control to constantly change

iewing a Web document with a Windows graphical client such as Microsoft's Internet
Explorer or Netscape's Navigator is something of a cross between reading a magazine
and watching television. Textual information is displayed with typographic fonts and
color graphics. The static display of information is often supplemented by animated
graphics and background music. Sound and video clips can be activated by clicking an
icon embedded in the document. Clicking on a highlighted word or phrase in the

may cause the reader to skip to another part of the displayed document, or it
may cause yet another document to be retrieved. The Web leads the reader to skip from
one document to another, retrieving information from servers scattered around the world.


lectronic mail probably is still the most widely used Internet function. A commonly
used configuration requires that a user have an account on a POP (Post Office Protocol)
mail server. The e
mail client software accesses the server and downloads any

messages to the user's PC. Mail composed at the user's PC is transmitted to the server,
from where it is sent to its ultimate destination anywhere in the world.


nternet news, also referred to as USENET news, is a conferencing syste
m made up of
tens of thousands of topical conferences known as news groups. Those familiar with the
electronic bulletin board systems of the 1980's and early 1990's will compare Internet
news to echo conferences. Others will draw an analogy to mailing list
s such as listserv on
BITNET. The user reads the news by using client software to subscribe to a selection of
news groups. When the client software accesses an NNTP (
rotocol) server, the server downloads to the client a list of subje
cts for all unread
messages stored on the server for the selected news group. The user can then select any
message for reading, post a response to the message to the group, or reply directly to the
original poster of the message. The client software mainta
ins on the user's PC a list of all
available groups on the server, along with records of which messages have been read or
skipped over. Only the messages selected for reading are actually downloaded to the
user's PC.


FTP (File Transfer Protocol) allow
s the transfer of files between any two computers of
any type. Files can be transferred from PC to PC, PC to mainframe, PC to Mac, PC to
UNIX machine, and vice versa. Any kind of computer file, whether it be a text file or a
binary file representing softwa
re, graphics images, or sounds, can be transferred. Of
course, whether the file is usable on the receiving machine depends on the nature of the
file and the availability of software to make use of the file.


elnet enables the user of a PC to login
to a host computer at another site on the Internet.
The user's PC then acts as a dumb terminal attached to the remote host. Such access
usually requires that the user have an account on the remote host. For instance, a student
or faculty member at one univ
ersity might have an account on a computer located at
another university. Many commercial database services that require only a character
based interface are available via telnet, including services such as the Dow Jones News
Service and the Lexis/Nexis se
rvice. Some services are available without charge. For
example, hundreds of libraries in all parts of the world allow free remote access to their
computerized catalogs and to some specialized databases.


opher was an early system that enabled the u
ser to find files and other Internet services
by navigating a system of text menus and submenus. As a corollary, it provided a means
for information providers to publish information on the Internet in a discoverable manner.
Prior to the development of Goph
er at the University of Minnesota in the early 1990's,
information on the Internet was located by asking friends and strangers where to look.
Gopher was rendered obsolete by the greater capabilities of the World Wide Web.
Despite Gopher's obsolescence, in
May 2000 a search on Yahoo for the phrase "gopher
server" returned more than 24,000 web pages, including some links to Gopher servers
that were still functioning.

he first step in using a Gopher client was to "point" the client at the address of a known

Gopher server. The client then retrieved that Gopher's menu of topics. Typically, many of
the topics on a Gopher menu were pointers to yet other menu items on other Gopher
servers. The fact that each item in the sequence of selections might have come from

different Gopher servers in widely scattered parts of the world was transparent to the
user. The Gopher client software presented the many different Gopher servers as if they
represented a single application on a single machine. Navigating such menus coul
d lead
the user to skip from one Gopher server to another, literally retrieving information from
servers scattered around the world in just a few minutes.

tems on Gopher menus could be of many different data types in addition to menus
listing choices of
topics. When an item such as a text, graphics, or sound file was
selected, the Gopher client transfered the file to the user's PC. Then, as an option, it might
load the file into an appropriate "viewer" selected by the user. A simple text file could be
ded into Windows Notepad. A graphics file in GIF or JPEG format might be loaded
into LVIEW, a popular shareware graphics viewer for Windows. A binary program file
would simply be downloaded into a designated directory for use at some other time.
Finding re
levant Gopher menu items was facilitated through the use of Veronica, which
was a database of the text of Gopher menus. Most Gopher servers included Veronica
access as a menu selection.

Windows and TCP/IP for Internet Access


t is helpfu
l to know some Internet terminology when working with your local network
specialist or Internet service provider to make your PC a host on the Internet. The two
common modes of Internet access are through a direct network connection or through a
serial con
nection to a SLIP or PPP server.


direct network connection involves installing a network interface card (NIC) in your
PC. Most likely this will be an ethernet card. This card in turn is connected to your
organization's local area network. Wiring usually

consists of coaxial cable (as in thin
wire ethernet) or twisted pair telephone wiring (as in 10Base
T ethernet). The local
network in turn must be connected to the Internet, and it must be capable of handling
TCP/IP data packets.

rotocol) is the method by which data
on the Internet is divided into packets of bytes. Each packet is delimited with header
information that includes the destination address where the packet is to be routed when it
is transmitted ov
er the Internet. The local network and your PC may also be using other
network protocols simultaneously with TCP/IP. For instance, your PC may already be
connected to a network using Novell, LANtastic, or Windows network protocols. A
standard Microsoft pro
tocol for connecting machines running various versions of
Windows is known as NetBEUI (NetBIOS Extended User Interface).


everal layers of software are involved in implementing a direct network connection. The
ease with which these layers a
re installed and configured depends on the version of
Windows that you use.

Network Card Drivers

f the PC has a direct connection to the Internet, there must first be software to control
the network interface card. A method commonly used with Windows 3.1

is to install a
piece of software called a packet driver that deals directly with the network interface
card. The packet driver is loaded under DOS from the AUTOEXEC.BAT file as a TSR
(terminate and stay resident) program. A packet driver should be includ
ed with the
software that comes with the network interface card. If the manufacturer of the card does
not supply a packet driver, free packet drivers are available in the
Crynwr Pa
cket Driver
. A packet driver would not normally be used on machines running Windows
95 because Win95 includes its own 32
bit NDIS (
pecification) drivers for a wide variety of network interface cards.

Serial port driver

hen a direct network connection is not available, Internet TCP/IP software can be used
via phone lines to connect to a SLIP (Serial Line Internet Protocol) or PPP (Point to Point
Protocol) server that provides a connection to the Internet. SLIP and PPP
do not require a
packet driver since there is no network card involved. They do, however, require a SLIP
or PPP driver that communicates through the serial port. The Trumpet Winsock
shareware package and Windows 95 both include SLIP and PPP drivers. The TC
on package for Windows for Workgroups does not support serial communications,

he user first obtains an account on a SLIP or PPP server. Connecting to the Internet
involves dialing the server using normal serial communications software a
nd establishing
a SLIP or PPP session. Once the session is established, TCP/IP software running on the
PC can be used just as if the PC was connected directly to the Internet through a network
card. SLIP and PPP users are well advised to settle for nothing

less than transmission at
28,800 bits per second. World Wide Web especially transmits a great deal of data when
images or sound are involved. Slow modems and slow connections will discourage
anyone but the most dedicated user from exploring the possibilit
ies of the Internet.

TCP/IP protocol stack

he next layer of software is the TCP/IP driver, which can be implemented in a variety
of ways. In 1993, 16
bit TCP/IP was often implemented as another DOS TSR program
loaded from the AUTOEXEC.BAT file. Somewhat
later this layer of software was
implemented as a Windows dynamic link library (DLL) or virtual device driver (VxD).
The DLL and VxD implementations do not require any modification of the boot files on
the PC.

he TCP/IP driver, which implements TCP/IP fu
nctionality for the system, is referred to
as the TCP/IP protocol stack. This driver may be written to interface with a packet driver,
as is the case with the Trumpet Winsock package. The advantage of having a TCP/IP
stack that interfaces with a packet dri
ver is that a particular TCP/IP stack from one
vendor can be used with any network card for which an associated packet driver is
available. Thus, the packet driver specification eliminates the need for software vendors
to provide hardware drivers for every

possible network card that might be used with their
TCP/IP stack. When using a packet driver with Windows 3.x applications, another DOS
TSR referred to as a virtual packet driver may be required to interface between the
based TCP/IP protocol stack

and the DOS
based packet driver.

he TCP/IP protocol stack included with Windows 95 does not use a packet driver
interface with the network card. Instead, Windows 95 provides drivers for most popular
network cards. Hardware vendors now provide Windows 95

drivers for their network
cards, just as they provided packet drivers in the past. Although packet driver based
TCP/IP stacks from other vendors can be used with Windows 95, it is far preferable to
use the 32
bit TCP/IP support integrated into Windows 95.

Client applications

TCP/IP client applications work at the top of the layers of software so far described.
Client software runs independently of the type of connection to the Internet. TCP/IP
applications frequently are referred to as clients because the
y access a corresponding
server (a daemon in UNIX terminology) on another machine. An FTP client, for instance,
is the application on the user's machine that accesses the FTP server running on a host
computer located elsewhere on the Internet.

ntil about

1993, each TCP/IP client had to be written to interface with a particular
vendor's TCP/IP protocol stack. Clients that worked with one vendor's TCP/IP driver
would not work with a driver from another vendor. This restriction was eliminated with
the develo
pment of the Windows Sockets Application Programming Interface, otherwise
known as the Winsock API, or more simply Winsock. Winsock works in the layer
between the TCP/IP client and the TCP/IP protocol stack. Winsock is the reason that all
the applications
that I used with Trumpet Winsock under Windows 3.1 worked just fine,
if not better, under the Windows 95 TCP/IP stack.

Windows and TCP/IP for Internet Access


s Internet connectivity has become more transparent to the user, the technical term and
popular buzzword "Winsock" is heard much less frequently than it was a few years ago.
Winsock technology is being absorbed into Windows and into applications software so
that end
users need not think about it very much. However, some understanding of the
echnology can be helpful in resolving problems that can still arise when one vendor's
Internet software overwrites files previously installed by another vendor's product.

indows TCP/IP applications usually involve Winsock technology. All of the client
tware described in this tutorial is based on Winsock.

insock (short for Windows sockets) is a technical specification that defines a standard
interface between a Windows TCP/IP client application (such as an e
mail client or an
FTP client) and the underl
ying TCP/IP protocol stack. The nomenclature is based on the
Sockets applications programming interface model used in Berkeley UNIX for
communications between programs.


Winsock compliant application like Eudora get's its networking capabilities by calli
standard procedures from a Windows dynamic link library. 16
bit TCP/IP applications
depend on a file named WINSOCK.DLL. 32
bit applications written for use with
Windows 95 or Windows NT depend on a file named WSOCK32.DLL. Procedures
called from these Wi
nsock files in turn invoke procedures in the hardware drivers
supplied with the TCP/IP protocol stack. As described earlier, the TCP/IP driver
communicates with the computer's ethernet card, either directly or through a packet

he files WINSOCK.DL
L and WSOCK32.DLL are not generic files that can be freely
exchanged among computers running TCP/IP protocol stacks from different vendors.
Each vendor of a TCP/IP protocol stack supplies a proprietary WINSOCK.DLL and
WSOCK32.DLL that works only with that
vendor's TCP/IP stack.

evelopers of Internet applications write their software so that it conforms to the
Winsock standard. The advantage to the developer of the software is that the application
will work with any vendor's Winsock implementation. Thus, th
e developer of an
application such as an FTP client has to understand the Winsock interface, but he does
not have to know the details of each vendor's TCP/IP protocol stack in order to make his
client application compatible with that stack.

insock also el
iminates the need for an application developer to include his own TCP/IP
protocol stack within the application program itself. It perhaps seems unbelievable today,
but this was a common means of implementing TCP/IP clients under DOS. Some early
Windows TCP
/IP clients also used this method. The use of protocol stacks internal to the
client results in conflicts when two clients try to access the single packet driver that is
communicating with the network card. The ability to create applications compatible wit
any vendor's Winsock compliant protocol stack resulted in a blossoming of Winsock
compliant shareware applications in summer 1993.

he Winsock standard offers advantages to the end
user. One advantage is that several
Winsock applications from different
vendors can be used simultaneously. This is a
marked improvement over earlier packet driver applications in which each application
contained a built
in TCP/IP stack. Such applications cannot share the packet driver
except through the added complexity of a
packet multiplexer such as PKTMUX. A
second advantage to the user is that any Winsock compliant application will run with any
vendor's TCP/IP protocol stack and accompanying WINSOCK.DLL.

s recently as 1995, some commercial vendors of TCP/IP clients were
not taking
advantage of Winsock capabilities. There were TCP/IP clients that required dedicated
access to the packet driver, and there were clients that would run only with the TCP/IP
protocol stack supplied by one particular vendor. Fortunately, the unive
rsal adoption of
the Winsock standard has eliminated much of the complexity of communicating over the