Understand ...

The driving force behind enterprise networking is the shift toward an information-based business economy and the Internet.

Interconnecting the organization's diverse networks is one way to meet this challenge.

A backbone network is a high speed network that connects many networks in a single company or government site. It may also be called an enterprise network if it connects all networks within a company.

There are two approaches to providing high speed networking.

There are two basic components to a backbone network:

Operating at the physical layer, hubs are very simple devices that pass all traffic in both directions between the LAN sections they link.

They may connect different types of cable, but use the same data link and network protocol.

Strictly speaking, hubs are not considered part of a backbone network, but are usually repeaters or amplifiers.

Bridges operate at the data link layer. They connect two LAN segments that use the same data link and network protocol. They may use the same or different types of cables.

Bridges "learn" whether to forward packets, and only forward those messages that need to go to other network segments.

If a bridge receives a packet with a destination address that is not in the address table, it forwards the packet to all networks or network segments except the one on which it was received.

Bridges are a combination of both hardware and software, typically a "black box" that sits between the two networks, but can also be a computer with two NICs and special software.

Like bridges, switches operate at the data link layer. Switches connect two or more computers or network segments that use the same data link and network protocol. They may connect the same or different types of cable.

Switches operate at the same layers as bridges but differ from them in two ways:

There are two types of switches:

Cut-through switches examine the destination of the incoming packet and immediately connect the port with the incoming message to the correct outgoing port.

Store-and-forward switches copy the incoming packet into memory before processing the destination address.

Routers operate at the network layer. Routers connect two or more LANs that use the same or different data link protocols, but the same network protocol.

Routers may be "black boxes," computers with several NICs, or special network modules in computers.

In general they perform more processing on each message than bridges and therefore operate more slowly.

Routers choose the best route between networks when there are several possible routes between them.

Routers also only process messages specifically addressed to it, unlike bridges.

Routers make no changes to the network layer packet and user data it receives.

Brouters are devices that combine the functions of both bridges and routers. These operate at both the data link and network layers. A brouter connects both same and different data link type network LAN segments.

It is as fast as a bridge for same data link type networks, but can also connect different data link type networks.

Gateways operate at the network layer and use network layer addresses in processing messages.

Gateways connect two or more LANs that use the same or different (usually different) data link and network protocols. The may connect the same or different kings of cable.

Gateways process only those messages explicitly addressed to them.

Gateways translate one network protocol into another, translate data formats, and open sessions between application programs, thus overcoming both hardware and software incompatibilities.

A gateway may be a stand-alone microcomputer with several NICs and special software, a FEP connected to a mainframe computer, or even a special circuit card in the network server.

One of the most common uses of gateways is to enable LANs that use TCP/IP and ethernet to communicate with IBM mainframes that use SNA.

The gateway provides both the basic system interconnection and the necessary translation between the protocols in both directions.

One approach to providing high speed networks is to develop faster versions of the same technology used in LANs.

The technologies operate by providing a high speed circuit that is shared among all computers on the LAN or backbone network.

The concept behind fast ethernet is simple: take an ethernet LAN and make it run faster.

There are several fundamentally different approaches in the marketplace: Those that refer to 100Mbps "Fast Ethernet" and "faster" ethernet: gigabit ethernet and Iso-ENET.

100Base-X, (a.k.a. 100Base-T), is virtually identical to 10Base-T (IEEE 802.3). It gives a 100 Mbps data rate using the standard ethernet bus topology, data link packets and CSMA/CD media access protocol.

There are three versions of 100Base-X that differ only at the physical layer:

100Base-VG (a.k.a. 100VG-AnyLAN) is considered part of the fast ethernet family but is really not ethernet.

There are two differences between 100Base-VG and 100Base-X:

In theory 100Base-VG should be faster than 100Base-T when network traffic becomes heavy because controlled media access techniques such as DPAM perform better than contention-based approaches like CSMA/CD.

When traffic is heavy, the throughput of ethernet drops to about 50 % because of the collisions that occur. For larger packets this is also true for 100Base-T. For small packets of 1K or less, 100Bast-T operates at almost 100 Mbps. For large packets, throughput drops to about 50Mbps.

100Base-VG operates at close to 100Mbps regardless.

Similar to 100Base-X, 1000Base-X is a set of standards that provide 1 Gbps. One problem with 1000Base-X is that using the standard CSMA/CD media access control on a shared network may cause problems.

For this reason, gigabit ethernet may remain primarily a backbone technology for use only in point-to-point full duplex data communications links.

Four versions of 1000Base-X are now in use, and they differ only at the physical layer by using different media:

Iso-ENET (Isochronous ethernet) is standard 10BaseT ethernet with an additional 6.144 Mbps circuit placed on top. The two circuits are carried on the same physical UTP, but are completely separate to prevent interference.

Several vendors have developed desktop videoconferencing application software that uses iso-ENET with several more multimedia products under development.

Fast token ring (a.k.a. High Speed Token Ring (HSTR)) is similar in concept to fast ethernet. Fast token ring uses the standard token ring topology, protocols, and media access control, but runs at 100Mbps,and will run on cat 5 cable or fiber optic.

Sales of token ring have declined in recent years as more and more networks have switched to ethernet, which remains cheaper.

Fiber Distributed Data Interface (FDDI) is a set of standards originally designed in the late 1980s for use in MANs (ANSI X3T9.5), but has since made its way into backbone networks.

FDDI is a token-passing ring network that operates at 100 Mbps over two-counter-rotating fiber optic cable rings.

The FDDI standard assumes a maximum of 1000 stations and a 200k path that requires a repeater every 2k. The second ring is for backup.

Single attachment stations (SAS) and dual-attachment stations (DAS) are both computer that can connect to one or both of the rings, respectively.

If the cable in the FDDI ring is broken, the ring can still operate in a limited fashion.

The FDDI-MAC scheme uses a variation of the IEEE 802.5 token-passing standard used for token-ring networks with two types of tokens: free tokens and busy tokens.

When a computer on an FDDI network receives the token with a message attached, it first removes the token from the ring, and then transmits all messages that were attached to it. The computer then transmits whatever messages its wants before transmitting the token.

There are two additional types of FDDI:

Over the past few years, there has been a major change in the way we think about LANs and backbone networks. LANs have traditionally used multipoint circuits, and WANs have traditionally used point-to-point circuits.

As the shared circuits in LANs and BNs have become overloaded with traffic, networks are starting to use switched point-to-point circuits rather than shared multipoint circuits.

The concept behind switched ethernet - and all switched media technologies - is simple; replace the LAN hub with a switch. Each computer now has its own dedicated point-to-point circuit.

Switched ethernet dramatically improves LAN performance. However, since much of the network traffic is to and from the server, the circuit to the server is often the network bottleneck.

One obvious solution is to increase the number of connections from the server to the switch so that traffic now can reach the server on several circuits.

Other solutions include:

Switched FDDI is similar to switched ethernet, in that a FDDI switch replaces the FDDI hub, providing a series of point-to-point connections from the computers to the switch instead of the traditional shared circuit. The network has a star topology instead of a ring, and no token.

It does use the FDDI packet format and is fully compatible with other FDDI hardware.

Asynchronous Transfer Mode (ATM) (a.k.a. cell relay) is a technology originally designed for use in wide area networks that is now often used in backbone networks.

ATM backbone switches typically provide point-to-point full duplex circuits at 155 Mbps (total of 310 Mbps).

ATM is a switched network but differs from switched ethernet and switched token ring in four ways:

Asynchronous Transfer Mode (ATM) is connection-oriented so all packets travel in order through the virtual circuit. A virtual circuit can either be a:

ATM uses a very different type of protocol than traditional LANs. It has a small 53-byte fixed length packet and is connection-oriented.Ethernet and token ring use larger variable length packets and are typically connectionless.

Translation must be done to enable the LAN packets to flow over the ATM backbones. There are two approaches LAN encapsulation (LANE) and Multiprotocol over ATM (MPOA).

Translating from ethernet or token ring into ATM is not simple.

First the ethernet address must be translated into an ATM virtual circuit identifier for the circuit that leads from the edge switch to the edge switch nearest the destination.

Once the virtual circuit address for the destination data link layer address has been found, it can be used to transmit the packet through the ATM backbone.

Once the virtual circuit is ready, the LAN packet is broken into the series of ATM cells, and transmitted over the ATM backbone using the ATM virtual circuit identifier.

Unfortunately this process can cause quite a delay (a reduction of 40 to 50 %).

Multiprotocol over ATM (MPOA) is an extension to LANE.

ATM-25 is a low speed version of ATM which provides point-to-point full duplex circuits at 25.6 Mbps in each direction. It is an adaptation of token ring that runs over cat 3 cable and can even use token ring hardware if modified.

ATM-51 is another version designed for the desktop allowing 51.84 Mbps from computers to the switch.

Both of these ATMs appear to be good choices for desktop connections when ATM backbone networks are used. However, industry has been very slow to accept either and have instead moved to fast ethernet which is both cheaper and faster.

Fiber channel is relatively new networking technology, although it has been used inside computer and disk storage devices for several years.

Fiber channel was originally designed to provide high speed transmission over fiber optic cable. The maximum data rate is 1.062 Gbps up to 10 k with higher rates under development.

Fiber channel is a switched technology like ATM, and like ATM, all circuits are full duplex and provide several classes of service.

Fiber channel uses variable length data link layer packets with a minimum of 2048 bytes of user data in each, and performs routing error control using CRC-32.

Improving the performance of backbone networks is similar to improving LAN performance. First find the bottleneck, then solve it, or move it somewhere else.

You can improve performance by improving the computers and other devices in the network, by upgrading the circuits between computers, and by changing the demand placed on the network.

Increase Computer and Device Performance

Increase Circuit Capacity

Reduce Network Demand

The primary functions of computers and devices in backbone networks are routing and protocol translations. They can be improved with a faster routing protocol.

Static routing is faster than dynamic, but can impair circuit performance in high traffic situations.

Many of the newer backbone technologies have standards that are not fully developed.

FDDI and ATM require the translation or encapsulation of ethernet and token ring packets before they can flow through the backbone.

Translating protocols typically requires more processing than encapsulation, so encapsulation can improve performance if the backbone devices are the bottleneck.

Most backbone devices are store and forward devices.

If network circuits are the bottleneck there are several options:

5 important factors to consider: