Home

 

Contents

 

Previous PageNext Page

 

Novell's Guide to Internet Access Solutions

Understanding the Difference between TCP/IP and IPX/SPX

Let's take a brief look at the TCP/IP protocol suite versus the IPX/SPX protocol suite, which will provide you with some comparison data for selecting a configuration design.

TCP/IP is a universal communications protocol that supports the communications system of the Internet, as well as many other internetwork environments (such as UNIX environments, and even the NetWare environment). TCP/IP was integrated into Berkeley Software Distribution (BSD) UNIX in 1982 and has since been considered an integral part of the UNIX operating system.

Since the Internet uses the TCP/IP protocol suite, you must also use the TCP/IP protocol suite to communicate across the Internet. NetWare, however, uses IPX/SPX as its default communications suite. Figure 8.1 shows the TCP/IP protocol mapped to the OSI (Open Systems Interconnection) model. You'll notice that two of the layers, Session and Presentation, are missing from the model because the TCP/IP stack does not have protocols that map to those layers.

FIGURE 8.1 The TCP/IP protocol suite is mapped here to the OSI model.

As you can see in Figure 8.1, the TCP/IP protocol suite is comprised of a number of individual protocols. These protocols include:

  • IP. Used to route communications through an internetwork. Routing is based on the destination IP address.
  • TCP. Connection-oriented transport services for guaranteed delivery of data. HTTP uses TCP as its transport mechanism.
  • UDP. Connectionless transport services. Although much quicker than TCP, UDP does not guarantee delivery of data. An upper layer application must ensure the data was received.
  • ARP. Used to find a device's physical address based on its IP address. Stations transmit an ARP request to find a station's physical address and put the reply information in cache -- the ARP cache. The station does not need to send another ARP request if the entry is in cache.
  • RARP. Used to get an IP address when a station (such as a diskless workstation) only knows its physical address.

When your client communicates with a Web server host on the Internet using TCP/IP, the client makes an HTTP call that is packaged in an IP header. The header contains all of the destination address of the host Web server, as shown in Figure 8.2. The ISP's IP router forwards the packet onto the Internet based on the destination IP address.

FIGURE 8.2 The TCP/IP client places the HTTP request inside an IP header that is addressed to the Web server.

NetWare LANs, however, use IPX/SPX as their default communications protocol. These communications cannot traverse the Internet unless they are encapsulated in an IP header. Even then, the upper layer protocol, NCP, cannot be processed by the HTTP daemon. The NetWare client must either have a TCP/IP stack loaded (including the HTTP client software) or use a translational gateway between IPX and IP. The IPX protocol stack is shown in Figure 8.3.

FIGURE 8.3 The NetWare IPX/SPX protocol stack has three parts: IPX, SPX, and NCP.

NetWare's IPX/SPX protocol stack includes the following pieces:

  • IPX. Used to route communications through a NetWare internetwork. Routing is based on the destination IPX internetwork address. IPX provides connectionless transport services.
  • SPX. Connection-oriented transport services for guaranteed delivery of data.
  • NCP. Upper layer protocol that defines the function of the packet (file read, file write, login, logout, and so on).

NetWare communications are routed based on an IPX internetwork address -- an addressing scheme that can't be used on the Internet.

Copyright © 1996 Novell, Inc. All rights reserved.

 

Top of Page