With its open architecture, the PATHWORKS for OpenVMS Advanced Server software can operate over several popular protocols simultaneously, including:

• TCP/IP
• NetBEUI
• DECnet or DECnet-Plus (formerly DECnet/OSI)

This appendix provides information on the following topics:

• Understanding the OSI Reference Model --- describes the seven-layer networking software model.
• Choosing a Network Adapter Card --- provides pointers on how to select your network adapters.
• Choosing a Network Protocol --- briefly describes the protocols you can use, including TCP/IP, NetBEUI, and DECnet-Plus.

Before you explore the specific drivers and protocols supported by the PATHWORKS Advanced Server, you should understand both the OSI Reference Model and the purpose of network interface card drivers. If you already understand these topics, you can skip to Choosing a Network Protocol, which includes an overview and description of each protocol that interoperates with the PATHWORKS Advanced Server.

B.2 Understanding the OSI Reference Model

In 1978 the International Organization for Standardization (ISO) developed a model for computer networking called the Open Systems Interconnection (OSI) Reference Model. The model describes the flow of data in a computer network --- from the physical connections of the network to the applications used by the end user.

The OSI Reference Model is an idealized version of networking; few systems follow it exactly. However, the model is useful for discussion and comparison of networks.

The OSI Reference Model includes seven layers, as shown in the following figure. Each of the layers is responsible for a specific and discrete aspect of networking.

Figure B-1 OSI Reference Model

• The Physical Layer is responsible for getting bits from one computer to another. It also regulates the transmission of a stream of bits over a physical medium. This layer defines how the cable is attached to the network adapter card and which transmission technique is used to send data over the cable. It also defines bit synchronization and checking.
• The Data Link Layer packages raw bits from the Physical Layer into frames. A frame is a logical, structured packet in which data can be placed. The Data Link Layer is responsible for transferring frames from one computer to another without errors. After the Data Link Layer sends a frame, it waits for an acknowledgment from the receiving computer. Frames that are not acknowledged are resent.
• The Network Layer addresses messages and translates logical addresses and names into physical addresses. It also determines the route along the network from the source to the destination computer, and it manages traffic problems such as switching, routing, and controlling the congestion of data packets.
• The Transport Layer is responsible for error recognition and recovery, ensuring the reliable delivery of messages. It also repackages messages when necessary by dividing long messages into small packets for transmission. At the receiving end, it rebuilds the small packets into the original message. The receiving Transport Layer also sends an acknowledgment of receipt.
• The Session Layer allows two applications on different computers to establish, use, and end a session. This layer establishes dialog control between the two computers in a session, regulating which side transmits, when, and for how long.
• The Presentation Layer translates data from the Application Layer into an intermediary format. This layer also manages security issues by providing services such as data encryption, and it compresses data to reduce the number of bits that need to be transferred on the network.
• The Application Layer enables end-user applications to access network services.
  When two computers communicate over a network, the software at each layer on one computer communicates with the same layer on the other computer. For example, the Transport Layer of one computer communicates with the Transport Layer on the other computer. As shown in the following figure, the Transport Layer on the first computer is not involved with how the communication actually passes through the lower layers of the first computer, passes across the physical media, and up through the lower layers of the second computer.

Figure B-2 Transport Protocol

A network adapter card, also called a network interface card or a network interface controller (NIC), is an adapter board installed in a computer to let it function on a network. The network adapter card provides ports to which the network cable can connect physically. The card physically transmits data from the computer to the network cable, and back.

Every network computer must have a network adapter card driver, a software driver that controls the network card. Every network adapter card driver is configured to run with a certain type of network card.

When choosing network adapter cards, you first must choose cards that support your network's architecture (such as Ethernet or Token Ring) and cabling media (such as Thinnet or twisted pair). You also should consider the tradeoffs of performance and cost.

Performance for network adapter cards depends mostly on bus width and onboard memory. The best performance is achieved when the bus width of the card closely matches the internal bus width of the computer. Onboard memory enables a card to buffer frames going to and from the network. A card with the most memory is not always the best choice. At some point, diminishing returns and the maximum speed of other network components limit the performance gains of onboard memory.

When you consider the cost of network cards, factor in the cost of buying spare cards to replace the ones that fail. You should also ensure that your network hardware budget allows for cable, hubs, repeaters, routers, and other hardware, as well as the labor costs associated with installing them.

Before you decide on a type of network card, make sure that the OpenVMS operating system you are using supports it. Also, make sure the vendor can support your business needs. If you are working with a reseller, check that the reseller has good communication with the card manufacturer.

B.4 Choosing a Network Protocol

In addition to the network card and the network card driver, a network computer must have a protocol driver, also called a transport protocol or a protocol. The protocol driver works between the upper-level network software---such as the workstation and server---and the network adapter card. The protocol packages the data that are sent over the network in a way that the computer on the receiving end will understand.

The process of associating a protocol driver with the network adapter card with which it will work and establishing a communication channel between the two is called binding.

For two computers to communicate on a network, they must use identical protocols. In the case where computers are configured to use multiple protocols, they need to have only one protocol in common to communicate. For example, a server that uses both NetBEUI and TCP/IP can communicate both with workstations that use only NetBEUI and with workstations that use only TCP/IP.

The PATHWORKS Advanced Server allows connections from the transports and protocols shown in the following table.

Table B-1 Supported Transports and Protocols

Protocol	Client Transport	Server Transport Component
TCP/IP	Internet	Product-specific
NetBEUI (with NetBIOS)	LAN Manager	LAN Manager
DECnet (proprietary)	DECnet	DECnet

The remainder of this section provides an overview of each of these protocols with basic information about each protocol and its advantages and disadvantages.

B.4.1 TCP/IP Protocol

TCP/IP was developed in the late 1970s as a result of a research project on network interconnection by the Department of Defense Advanced Research Projects Agency (known as ARPANet, the precursor to the Internet). TCP/IP is actually a suite of protocols that defines various interactions between computers sharing the protocol.

Until recently, TCP/IP supported networking for midrange computers; however, as the PC began its rise in popularity, TCP/IP became a standard protocol for support in the PC environment.

TCP/IP has a reputation as a difficult protocol to configure and manage. However, new implementations are making it easier. For example, in TCP/IP, the Dynamic Host Configuration Protocol (DHCP) provides server support and is one of the most important advances in Windows 95 over Windows 3.x. Without DHCP, system administrators must manually assign the four-byte IP addresses to each computer. With DHCP enabled, a DHCP server can manage a range of IP addresses and assign one to each computer as it logs on to the network.

The principal advantage of TCP/IP is that it provides communication across interconnected networks with different operating systems and hardware architectures.

TCP/IP also provides compatibility with the Internet, a collection of networks and gateways linking universities, corporations, government offices, and military installations world wide.

The following table summarizes the advantages and disadvantages of using the TCP/IP protocol.

Table B-2 TCP/IP Protocol

Advantages	Disadvantages
Provides connectivity across different operating systems and hardware platforms.	Slower than NetBEUI on small LANs.
Provides internet connectivity.	Can be difficult to administer.
Provides routing support.	More overhead than NetBEUI.

B.4.2 NetBEUI Protocol

The NetBIOS Extended User Interface (NetBEUI) was first introduced by IBM in 1985. NetBIOS, an integral part of the NetBEUI protocol driver, is a programming interface that implements many session layer functions. NetBEUI is a small, efficient, and fast protocol with low overhead.

One reason for NetBEUI's lower overhead is that NetBEUI does not require an explicit acknowledgment (ACK) of each frame before it sends the next. Instead, the computer packages up several ACKs and sends them all at once. Requiring an ACK for every packet wastes network resources. NetBEUI dynamically determines the number of frames the sender can transmit before receiving an ACK, based on the network's current conditions.

NetBEUI was developed for LANs segmented into workgroups of 20 to 200 computers, with gateways connecting LAN segments to one another or to mainframes. NetBEUI is optimized for very high performance when used in departmental LANs or LAN segments. For traffic within a LAN segment, NetBEUI typically is the fastest protocol.

While NetBEUI is fast on small LANs, it is not so effective on large networks because it has a poor addressing scheme. NetBEUI does not allow duplicate computer names on the same network. This prevents a network from having two computers with the same name --- something difficult to eliminate on a large network.

The following table summarizes the advantages and disadvantages of the NetBEUI protocol.

Table B-3 NetBEUI Protocol

Advantages	Disadvantages
Tuned for small LAN communication, and therefore is very fast on LANs.	Not routable.
Good error protection.	Performance across WANs is poor.
Small memory usage.	Requires each network computer to have a unique name.

B.4.3 DECnet-Plus Protocol

DECnet-Plus is a proprietary protocol of Compaq Computer Corporation; it is a collection of many layered protocols offered together as a major data communications network. Developed as a distributed network, it supports a wide range of applications and programs.

One of the DECnet protocol's major advantages is flexibility in network configuration and applications functionality. DECnet-Plus includes the Local Area Transport (LAT) protocol that terminal servers use to communicate with hosts.

The following table lists the advantages and disadvantages of the DECnet-Plus protocol:

Table B-4 DECnet-Plus Protocol

Advantages	Disadvantages
Major flexibility in network configuration.	Complex network architecture.
Routable.	Proprietary.
Maintains a high level of availability, even in the event of node or link failure.
Supports a wide range of communications facilities, such as Ethernet and X.25.