 | TCP/IP History |
| TCP/IP Features |
| Internet Activities Board (IAB) |
| Request for Comments (RFC) |
| Network Information Center (NIC) |
| Management Services |
| Microsoft Implementation of TCP/IP |
This chapter covers the basics of how TCP/IP evolved and became the standard for the Internet and the worldwide networking industry. We discuss TCP/IP’s early development and the governing agencies that control how the Internet. is used. We also discuss the features of TCP/IP services that make it globally versatile and functional. Finally, we take a look at Microsoft’s redesigned and rewritten 32-bit TCP/IP implementation.
Well, you’ve finally gotten around to picking up a book that will teach you about TCP/IP. You can expect to learn the fundamentals of TCP/IP and gain an excellent working knowledge of this versatile network protocol. Before we begin, let’s take a look at the history of TCP/IP, the most popular networking protocol suite in the commercial and educational world!
TCP/IP was originally developed as a standard protocol to connect heterogeneous hardware and software operating systems in a network across cable or satellite, making it in effect a Wide Area Network (WAN). This network initially consisted of the University of Utah, the Stanford Research Institute, the University of California at Los Angeles, and the University of California at San Bernardino.
TCP/IP stands for Transmission Control Protocol (TCP) / Internet Protocol (IP). Don’t let its name fool you; TCP/IP is a suite of protocols, not just the two mentioned here. As we’ll see in the next chapter, each protocol in the suite has a specific purpose and function. Before we go any further let’s see how the world’s largest network, t Internet, became a reality.
Initially the Internet was known as the ARPANET.It started when the Department of Defense (DoD) needed to communicate across a Wide Area Network, but the current networks weren’t compatible or didn’t have cross–platform communication features. In addition to building a network that communicated across platforms, numerous services and functions had to be addressed, as we’ll see a little later in this chapter.
In the 1960s the U.S. Defense Advanced Research Projects Agency (DARPA) created and funded the development of a high-speed, packet switching network that would connect several research and government sites. ARPA was then renamed to DARPA. This network initially used the Xerox Networking System (XNS) protocol, but because of limitations XNS didn’t pan out. Soon a new transport protocol and routing protocol were developed, which became known as the TCP/IP Suite.
The DARPA project was funded by the Department of Defense and was headed by Dr. J.C.R. Licklider. Dr. Licklider gathered a group of computer specialists, nicknamed the "Intergalactic Network." Their goal was to develop the architecture needed for the Internet, which included the protocols and services that eventually became known as TCP/IP.
This architecture had to be modular and independent of hardware or software. In addition, the information on how to implement this network had to be public. It’s important to recognize that DARPA wasn’t responsible for building the Internet, but for designing the Internet model. This model also included defining the structure for the network protocols, which later became the TCP/IP protocol suite.
In 1967 DARPA contracted with the Stanford Research Institute to develop of the specification for this new communication system. In 1970 development began and by 1972 there were approximately 40 sites and support for TCP/IP utilities was begun. In 1973 the first international connection was established and 1974 released TCP/IP to the public.
The first implementation of this network consisted of government, military, and educational sites and it slowly grew to include commercial companies as well. This network has grown to enable communication throughout most of the countries in the world.
The Internet consists of many large networks that work together. If you have access from your PC to the Internet, you are using an Internet Service Provider (ISP). Your ISP provides your access to the Internet, which may include Web services. We’ll cover the Web in a later chapter.
In the late 1960s AT&T Bell Labs programmers Ken Thompson and Dennis Richards developed a new software operating system written entirely in C. This new operating system was called UNIX. In the early 70s AT&T gave the operating system code to universities for further development. To encourage the use of these internetworking protocols, DARPA made a low-cost implmentation available to UNIX developers and provided funds for its inclusion with the software distribution of UNIX. In 1980 the University of California at Berkeley integrated DARPA’s TCP/IP protocol suite as the networking component of UNIX. Hence, the marriage of UNIX and TCP/IP.
UNIX has continued to use TCP/IP as its network protocol and development has continued by networking vendors to add additional services to the TCP/IP suite of protocols and services.
The TCP/IP protocols were designed to meet the need for global communication. Let’s take a look at the features that make TCP/IP the standard for Internet connectivity.
Each layer in the TCP/IP protocol stack is modular, which means that it communicates with only the layer above or below it. So long as network developers write their device drivers or applications based on public Request for Comments (RFCs), there is no need to rewrite the entire protocol stack. The developer has to write only the code for the layer that his/her software will be implementing.
Because TCP/IP is not tied to an operating system, any vendor developing a new operating system with a network component can reference the RFCs to build a TCP/IP component. So a UNIX system will be able to talk to a VMS system, and the VMS system will be able to talk to an IBM mainframe, and the beat goes on. Even those operating systems with proprietary networks are jumping on the bandwagon. For example, Digital Equipment Corporation, Inc. has implemented TCP/IP services for OpenVMS.
TCP/IP allows any pair of processes running on different computers to communicate. This is most commonly used in a client/server environment. Here one process, typically the server, is residing on a central node and the client process will request certain tasks or information to be retrieved by the server.
TCP/IP’s unique addressing mechanism provides for over 4.2 billion addresses. Each host is referred to by its unique 32-bit address. These unique addresses are made up of network and host identification.
But with the growth of the Internet the Network Information Center (NIC) is already running out of addresses, and in fact since April 1995 has not assigned any new addresses. We will discuss a little later how to get an IP address and host name. The next implementation of TCP/IP and the Internet is in development now. IPV6 will use a 128-bit address, and hopefully this will last for a while!
The 32-bit IP address is broken into four octets that can be represented in decimal or binary format:
11010100 00001111 10000100 01110101 Binary Representation of Address
212.15.132.117 Dotted-Decimal Representation of the Address
Using the IP address scheme we further fragment this address into the network ID and the host ID. The Internet is one large communication network that consists of multiple networks. Each machine uses part of the IP address to identify the network it belongs to and the rest of the address to identify its host or local computer address. In Chapter 3 you will learn how to distinguish the network ID from the host ID and all their different implications.
A subnet mask is used to determine which part of the IP address is used for the network ID and which part for the host ID. There are two required parameters you must supply to initialize TCP/IP:
| IP address 4-octet address |
| Subnet mask 4-octet value |
As we just learned, the IP addressing scheme can build fairly large networks. Table 1-1 lists the three types of network addresses, called network classes, that are assigned to a company or an organization.
| Network Class | Number of Hosts |
| Class A | approximately 16,000,000 |
| Class B | approximately 64,000 |
| Class C | 254 |
Table 1 TCP/IP Network Classes and Number of Hosts
As you can see in Table 1-1, TCP/IP networks can become immense and still be able to transmit messages from one side of the world to another. In Chapter 3 we will examine all the possible combinations of addresses.
So how does a message move from one network to another?
We know that TCP/IP networks are made up of smaller networks either physically or logically segmented. To move between networks a router is used. A router is a computer that is connected to two or more networks. This computer forwards messages from one network to the other. In some TCP/IP implementations a router is also known as a gateway. A gateway is a computer that does protocol translation as well as routing.
For the large networks, network specialists use dedicated computers that perform nothing but routing operations. Bay Networks, Digital, 3Com, and Cisco are a few network vendors manufacturing large routers.
In small Microsoft networks, a Windows NT computer can hold two or more cards with different network addresses and act as a router. Each card can be assigned as many as five different addresses.
Microsoft’s 32-bit version of TCP/IP can communicate with multiple hardware and software platforms. Communication works across vendors because both computers have a TCP/IP stack that is written to RFCs. Microsoft’s TCP/IP has specifically been written to promote communication with multiple system types.
If you’re on the Internet, then you’re using TCP/IP. Internet hosts can be just about any system: Sun, IBM, VAX, ALPHA, and even Microsoft. As a user or even as a Network Specialist you don’t need to know which operating system is being used on the other end; you just need to make sure you’re using TCP/IP.
In Windows NT Server 3.5x and 4.0 support has been built in for Macintosh coexistence. As long as the Apple Macintosh continues to support TCP/IP, any Macintosh user will be able to FTP and send e-mail to a TCP/IP host.
IBM has implemented a TCP/IP stack. Therefore communication with IBM mainframes and Microsoft hosts works smoothly if both computers are using TCP/IP as a network protocol.
UNIX systems have been using TCP/IP since 1980. It’s easy to connect to a UNIX system and download files as long as you are using TCP/IP services.
Digital has had a history of building proprietary systems, but with TCP/IP services for OpenVMS, Digital has stepped into the foray. The earlier name for TCP/IP on a Digital platform was UNIX Connection or UCX.
Printers can also be connected to the Internet. Digital PrintServers andHewlett-Packard Laser Jets are two examples of printers that can be adapted or come with Network Adapter cards. These cards can be assigned IP addresses and then used from any type of TCP/IP host that supports network printing.
The coordinating committee for the Internet’s design, engineering, and management is known as the Internet Activities Board (IAB). Formed in the early 1980s, the IAB has the following mission:
| Manage the existing Internet |
| Maintain standards |
| Manage technical issues |
| Act as an internal liaison and representative for the Internet community |
| Help evolve the Internet into a large-scale, high-speed Network of the Future |
| Performstrategic planning for the Internet |
| Identify long-range problems and opportunities |
The IAB manages two subcommittees focused on engineering and research. Each subcommittee is led by a chairperson and guided by a steering committee. The subcommittees report to the IAB via the chairperson.
The IRTF promotes research in networking and the development of future Internet technology. There will be some overlap of research between the IRTF and the IETF, but this overlap is considered crucial to the cross-fertilization and technology transfer.
The IETF is responsible for the operation, management, and evolution of the Internet. The steering committee of the IETF is known as the Internet Engineering Steering Group (IESG). The IAB has delegated to the IESG the general responsibility for making the Internet work and the resolution of all short- and mid-range protocol and architectural issues.
The Networking Community votes for the IAB and its two subcommittee members, who serve a two-year term. inThere is representation by major firms specializing in network design, hardware and software development, as well as educational representatives from the most prestigious educational institutions in the world.
The IAB informs the networking community of its decisions by publishing Requests for Comments. An RFC is an invitation to develop standards, recommendations, or requirements for TCP/IP and the Internet. It is a method of asking for multiple solutions and choosing the best one. The IAB uses its steering committees to evaluate the RFC responses. The RFC is then published under one of five classifications:
| Required | Must be implemented |
| Recommended | Should be implemented and usually is |
| Elective | Optionally implemented and usually is not |
| Limited Use | Possible usage on some computer systems |
| Not Recommended | Not to be implemented |
An RFC passes through several phases of development. First, it is assigned a number. Then it passes through three different stages: a Proposed Standard, a Draft Standard, and an Internet Standard. The RFC number is never used again. If another version or a more efficient method of implementing a service or protocol is developed, another number is assigned. You can get the latest RFCs from Microsoft’s Web site. (See Appendix A for the latest RFCs.)
The Stanford Research Institute (SRI) maintains an index of RFCs that provides the number, title, authors, issue date, and the total number of hardcopy pages. It also lists the online format for the RFC. The index also notes whether an RFC is obsolete or updated by another RFC. You can download this index from:
FTP.NISC.SRI.COM
and get the file FYI/FYI-index.txt.
You can also retrieve the RFC index by sending e-mail to:
mail-server@nisc.sri.com. In the body of the message place the following text:
| RFCs | FYIs |
| send RFC-index | send FYI-index |
You can download RFCs from the following FTP sites:
| NIS.NSF.NET |
| NISC.JVNC.NET |
| VENERA.ISI.EDU |
| WUARCHIVE.WUSTL.EDU |
| SRC.DOC.IC.AC.UK |
| FTP.CONCERT.NET |
| DS.INTERNIC.NET |
| NIC.DDN.MIL |
For specific help in retrieving RFCs from these sites, send an e-mail message to rfc-info@ISI.EDU with the following in the body of the message:
help: ways_to_get_rfcs
The IAB is essentially responsible for assigning IP addresses to individuals, companies, organizations, and corporations on the Internet. The IAB has given the Network Information Center (NIC) the responsibility for assigning valid TCP/IP addresses. The only problem is that the NIC has been out of addresses since April 1995. You can obtain an IP address only through an ISP.
You ISP will assign you an IP address and domain name. You are then responsible for maintaining theconnection to your ISP, which may include space for Web pages.
IP addresses are hard for humans to remember. The Domain Name Service (DNS) was created so that instead of having to type the IP address of a company’s Web site, for example, you can access that site by a user-friendly name.
This service is based on a hierarchical structure that is read from right to left as you move down through the hosts. When the left-most name of an address is reached, the IP address of the host should be returned. You use domain names every time you search on the Web or send e-mail. Here are some DNS examples:
| Type of Address | Example |
| Web-based address | www.syngress.com |
| E-mail address | RuddyM@globalknowledge.com |
It’s important to contact your ISP and verify that your host name, IP address, and domain name do not conflict with other hosts on their network.
Microsoft’s TCP/IP services have definitely come of age. In the first version of NT some of these services needed just a little more work to even out the rough edges. With the release of Windows NT Server 4.0, it looks like they’re finally there!
This service provides PCs with automatic configuration of the three necessary TCP/IP parameters: IP address, subnet mask, and default gateway.
The first step in a DHCP environment is to identify a server as a DHCP server. This server maintains a pool of addresses and maintains the assignment of these addresses to the requesting computers. During boot-up, requesting computers issue a request for IP parameters. The DHCP server then responds to the requesting computer with the appropriate parameters and maintains its assignment in the DHCP database.
The automatic assignment of IP parameters makes configuration of a large number of computers painless. DHCP also provides automation, reducing human error. (DHCP is discussed in Chapter n.)
The second of many solutions to host name resolution is Microsoft’s Windows Internet Name Service. This service provides a NetBIOS-computer-name to IP-address mapping in a database on a server. The main feature of WINS is that instead of having to maintain a separate LMHOST file for NetBIOS computer names on each separate computer, a large database can be maintained on a server that keeps track of IP to NetBIOS name mappings. The WINS server database is dynamic, unlike the LMHOSTS files that must be modified manually.
Whenever a NetBIOS computer needs to resolve a NetBIOS name to an IP address, the requesting computer contacts the WINS server for a name resolution. The server then responds with the latest known IP address of the requested client.
Microsoft’s implementation of DNS is finally ready to graduate. Windows NT version 3.5 did not fully support DNS, but in Windows NT 4.0 DNS is ready to roll. A Windows NT 4.0 server can be configured as a DNS server.
DNS provides a database and search algorithm for resolving the host name into an IP address. DNS is the service that takes a domain name and resolves it to an IP address for TCP/IP hosts. This is not the same as the WINS server, which resolves NetBIOS computer names. UNIX hosts do not use NetBIOS names; neither do OpenVMS hosts. The DNS server eliminates the need for HOSTS file but this database is not dynamic; therefore all changes must be made manually.
DNS provides a hierarchical naming structure. At the top of the hierarchy is the root domain where all the top-level domains reside. You have probably used a top-level domain name at one time or another, such as .org, .com, .net, or .edu.
You can tell immediately which domain type you’re working with by the last domain specification, as shown in Table 1-2.
| Domain Name | Organization |
| COM | Commercial organization |
| EDU | Educational institution |
| GOV | Government organization |
| MIL | Military |
| NET | Network support provider |
| ORG | Non-profit or other than above |
| INT | International organization |
Table 2: We need a table title here. amw The Seven Domain Names.
The word to the left of the domain name is usually the company or organization name; for example: microsoft.com, sri.net, unitedway.org, and stanford.edu. These intuitive names are much easier to remember than a series of numbers.
Microsoft provides an SNMP agent that will track specific TCP/IP events. The SNMP server identifies which events to track using a database structure called a Management Information Database (MIB). When started, the SNMP service collects and delivers these events to the SNMP server. As of this writing, Microsoft has provided the SNMP agent and service on Windows NT 4.0, but not an SNMP server bundled in as a part of the distribution.
Microsoft’s TCP/IP suite for Windows NT 4.0 was designed for enterprise environments of large government and corporate sites. In addition, it was designed and written to be Internet-ready. This means that when you install Microsoft’s TCP/IP component, it is ready for connection to the Internet using Microsoft’s Internet products and features.
There have also been some performance enhancements that have made TCP/IP for Windows NT 4.0 more effective than the stack used previously by Microsoft in LAN Manager or Windows for Workgroups. Specific features in this new TCP/IP protocol suite are:
| Standards compliant |
| Interoperable |
| Portable |
| Scaleable |
| High performance |
| Versatile |
| Self-tuning |
| Easy to administer |
| Adaptable |
This is a new TCP/IP stack that Microsoft is bundling in with NT 4.0.
Microsoft has taken the design specifications in RFCs and redesigned and rewritten their TCP/IP stack. They have included many services that were unavailable in earlier versions:
| DHCP server and client |
| WINS server |
| DNS |
| PPP/SLIP dialup support |
| Point-to-Point Tunneling Protocol (PPTP) |
| TCP/IP network printing |
| SNMP agent |
| NetBIOS interface |
| Windows socket interface |
| Remote Procedure Call (RPC) |
| Network Dynamic Data Exchange (NetDDE) |
| Wide Area Network support |
| Basic TCP/IP utilities |
| Enhanced server software for simple network protocols |
| TCP/IP management and diagnostic tools |
Microsoft has implemented TCP/IP server features in both the Windows NT and the LAN manager line of products.
Windows NT 4.0 contains the newly rewritten Microsoft TCP/IP protocol stack that implements all of the features we’ve discussed.
The Microsoft LAN manager was developed when the old TCP/IP protocol stack was used. It contained code that was not specifically implemented to support large enterprises.
However, all the newest Microsoft products (Windows NT, Windows 95 and 98, and TCP/IP-32) share the new TCP/IP protocol suite described in this book.
Microsoft products acting in the client role were also implemented with the rewritten, redesigned TCP/IP code.
The Windows NT workstation contains all the basic client services code. Windows NT Workstation 3.5x can be designated as a DHCP, WINS, or DNS client and can use basic utilities such as FTP, Telnet, and the majority of the TCP/IP client utilities.
Windows 95 has also implemented the new Microsoft TCP/IP protocol stack. This client has the capability to become a DHCP, WINS, or DNS client and can also use the basic utilities of TCP/IP.
The new Microsoft TCP/IP protocol stack was not released in time to be included with Windows for Workgroups. It is distributed as an add-on-networking component and is available from the Microsoft Web site: www.microsoft.com.www.microsoft.com
We’ve said that Microsoft’s new TCP/IP protocol stack is Internet-ready. Let’s discuss what that means to you as a Network Specialist.
Microsoft has created two new products that complement the TCP/IP protocol suite and provide specific features for an Internet or Intranet.
The Internet Information Server (IIS) is released as part of the Windows NT 4.0 distribution. It contains the ability to configure and maintain the following types of Internet services:
| FTP (File Transfer Protocol) Users connect to this service and copy files to and from the site. Files can be stored in a single directory or a group of directories. |
| Gopher Gopher services can also be used for copying files. Although not widely in use today, support is still available. |
| World Wide Web (WWW) This is the Internet service used to host Web pages written in HyperText Markup Language (HTML). Pages are viewed with a Web browser using HyperText Transfer Protocol (HTTP). |
These repositories can then be secured using Microsoft’s NTFS file system security. In addition, each Internet service can be separately maintained or not at all. As a Administrator you can control the number of visits allowed at one time, or whether a user must log in to a valid NT account or use a guest account.
The second of Microsoft’s Internet-ready products is the Internet Proxy Server, which is a firewall and Web cache server that installs on the Windows NT 4.0 Server platform. The Proxy Server acts as a gateway with firewall protection between your Intranet and the Internet. Proxy Server 2.0 includes the following features:
| Dynamic packet filtering |
| Multilayered security |
| Altering and logging |
| Shielding of internal network addresses and Internet server applications |
| Virtual hosting |
We’ve taken a look at TCP/IP and its history, from its inception as part of the DARPA Project to its evolution as the protocol stack of choice for the Internet and Intranets.)
TCP/IP’s main features are its modularity, which is defined as the ability to write device drivers or application for a specific portion of the protocol stack without having to write the other protocol layer components. Using the TCP/IP 32-bit addressing scheme, large enterprise-size networks can be built and are fully functional. In addition, TCP/IP allows you interconnection with any vendor software operating system as long as physical connectivity exists.
Request for Comments are proposals for recommendations, standards, or requirements. RFCs are reviewed by the IAB and its two subcommittees. When an RFC is approved, it is assigned a number and then is classified as Required, Recommended, Elective, Limited Use, or Not Recommended. Copies of RFCs can be retrieved from several Internet locations.
The Network Information Center is responsible for the assignment of TCP/IP addresses, host names, and domain names on the Internet. The NIC has not given out any new addresses since April 1995 and you must contact an Internet Service Provider to connect to the Internet and be assigned an address.
TCP/IP management services such as DHCP, WINS, and DNS have been developed to make the configuration, maintenance, and management of TCP/IP easier.
Microsoft’s redesigned, rewritten 32-bit TCP/IP protocol stack includes new features and services for Windows NT 4.0 Server and Workstation, Windows 95 and 98, as well as TCP/IP-32 for Windows for Workgroups.
| TCP/IP is the most popular networking protocol suite in the commercial and educational world. |
| Be sure you understand the TCP/IP acronym. You will be tested on it. |
| TCP/IP was developed by the government to build a heterogeneous network across a wide area. |
| The TCP/IP protocols were designed to meet the need for global communication. |
| Each layer in the TCP/IP protocol stack is modular, which means that it communicates with only the layer above or below it. |
| Because TCP/IP is not tied to an operating system, any vendor developing a new operating system with a network component can reference the RFCs to build a TCP/IP component. |
| TCP/IP’s unique addressing mechanism provides for over 4.2 billion addresses. Each host is referred to by its unique 32-bit address. |
| Remember that the IP address is a 32-bit decimal address, separated into four equal octets by a ".". This number is then converted to binary and used as a unique identifier. |
| In some TCP/IP implementations a router is also known as a gateway. |
| Routing is the action of forwarding data between subnetworks. |
| The number of cards that can be placed into an NT server is dependent on whether the hardware has free resources and a free slot. The number of IP addresses assigned to a card is five. |
| Microsoft’s 32-bit version of TCP/IP can communicate with multiple hardware and software platforms. |
| The coordinating committee for the Internet’s design, engineering, and management is known as the Internet Activities Board (IAB). |
| Know the responsibiliteis of the IAB. |
| An RFC is an invitation to develop standards, recommendations, or requirements for TCP/IP and the Internet. |
| The Stanford Research Institute (SRI) maintains an index of RFCs that provides the number, title, authors, issue date, and the total number of hardcopy pages. |
| Your ISP will assign you an IP address and domain name. |
| The Domain Name Service (DNS) was created so that instead of having to type the IP address of a company’s Web site, you can access that site by a user-friendly name. |
| The DHCP service provides PCs with automatic configuration of the three necessary TCP/IP parameters: IP address, subnet mask, and default gateway. |
| Microsoft’s Windows Internet Name Service provides a NetBIOS-computer-name to IP-address mapping in a database on a server. |
| A Windows NT 4.0 server can be configured as a DNS server. DNS provides a database and search algorithm for resolving the host name into an IP address. |
| Microsoft provides an SNMP agent that will track specific TCP/IP events. The SNMP server identifies which events to track using a database structure called a Management Information Database (MIB). |
| When you install Microsoft’s TCP/IP component, it is ready for connection to the Internet using Microsoft’s Internet products and features. |
| The Internet Information Server (IIS) is released as part of the Windows NT 4.0 distribution. |
| The Internet Proxy Server is a firewall and Web cache server that installs on the Windows NT 4.0 Server platform. |