The chapter starts out explaining how a BN (Backbone Network) works. Backbone Network: Connecting all of the LANs of an organization entails another type of network (BN). A properly designed backbone network provides a high-speed circuit that serves as the central conduit across which the LANs of an organization can communicate.
They can also be used to connect LANs within a building, across a campus, and, increasingly, across much greater distances. A BN, as indicated by its name, is a network of its own. Besides connecting the various network segments, the backbone may have its own devices that can be accessed by other network segments Metropolitan area network (MAN): MAN spans a city and is often used to connect remote BNs. MAN in some cases can be considered a citywide BN and as the geographic distances they cover have increased, especially with the use of fiber-optics. BNs connect networks between floors of a building, across a city, or between states and countries. BN and MAN are sometimes used interchangeably, based on the scope of the BN. Network Segments: Each individual LAN owned by an organization is reffered to as a network segment.
Horizontal Segment: A moderate- to large-scale organization might have a network segment on each floor of a multistory building. Because each network segment, or LAN, typically occupies its own floor, this type of network segment is often referred to as a horizontal network. For example, assume that a business occupies three floors of a building. On each floor is a separate LAN, or horizontal network segment. Vertical Network: LANs could, and probably would, be connected to each other by a BN. This type of multi-floor connection is an example of a vertical network.
The BN in this instance is the central connecting cable running vertically from floor to floor that enables the horizontal networks to communicate with each other. Part of configuration analysis includes determining how each network segment connects to the BN. Generally, each network segment is connected to the BN using either a switch or a router. Deciding which Backbone Protocol to use? Questions to ask: 1) Traffic Demands
2) Constant Communication
3) Mostly Independent
Gigabyte Ethernet: is a very popular choice for BNs. The IEEE’s initial standard for Gigabit Ethernet is the 802.3z standard. Gigabit Ethernet allows for a data rate of 1,000 Mbps, or 1 Gbps. A major advantage of all of the officially recognized forms of Gigabit Ethernet is that each form builds on the standards of the preexisting Ethernet protocol. This means that the MAC layer and access method for Gigabit Ethernet are the same as those for standard and Fast Ethernet. Additionally, Gigabit Ethernet supports both half- and full-duplex communications. Other protocols that might be used for a backbone include Frame Relay (FR) and Asynchronous Transfer Mode (ATM).
Frame Relay and ATM are also frequently associated with WANs. As such, discussion of Frame Relay and ATM will be reserved for Chapter 7, which focuses on WANs. Backbone Architecture: two most common BN architectures are distributed and collapsed. Factors that influence a business’s decision as to which architecture to use include business needs, the condition of the physical facility (sometimes called the plant or campus), how users need to communicate, and the budget. The larger and more complex the organization, the more critical the decision becomes as to what type of backbone architecture to use. It can be very costly to change an existing backbone architecture once one has been put in place.
Distributed Backbone: runs throughout the entire enterprise. This type of backbone uses a central cable to which the network segments are connected. The central cable, which is the backbone, requires its own protocol, such as Gigabit Ethernet; it is also its own network. The backbone is considered to be distributed because each network segment has its own cabled connection to the backbone. The backbone is distributed to the LANs by connecting the LANs to the backbone. They maybe connected with routers and switches. In some cases even servers. A distributed backbone typically has separate routers that connect each logical network to the backbone. Because separate routers are used, internetwork traffic may have to pass through several routers to reach its destination. One advantage of a distributed backbone is that it allows resources required by most, if not all, internetworking users to be placed directly on the BN. Collapsed Backbone: connects all of the network segments to a central, single router or switch.
This central device is, in effect, the backbone. The network segments typically connect to the central backbone device by means of a hub, switch, or router. Because only a single,central backbone device is used, cabling is greatly reduced. Furthermore, additional connecting devices are not required. A collapsed backbone can result in significant cost savings. Backplane: is an internal, high-speed communications bus that is used in place of the connecting cables found in a distributed backbone. Because fiber-optic cabling is used to connect network segments to the collapsed backbone’s backplane, long distances are possible. With fiber-optic cabling, network segments may be widely scattered across a building or even a campus.
Backbone Fault Tolerance: is the capability of a technology to recover in the event of error, failure, or some other unexpected event that disrupts organizational communications and functions. Should the backbone fail for some reason, internetworking may no longer be possible. In such an event, business could come to a standstill and, depending on the recovery time, irreparable damage may occur. However, if fault tolerance has been built into the backbone, internetworking will likely still be possible. Fault tolerance will determine its ability to survive an error, damage, or some other unforeseen circumstance. Redundant Backbone: Should one backbone become unavailable, the other can still be used for internetworking traffic.
Furthermore, using a redundant backbone also allows for the load balancing of internetworking traffic. By placing half of the network segments on each backbone, internetworking traffic is shared, or balanced, across the backbones, resulting in improved communications performance. It is also VERY expensive. Wiring Closet: The patch panel is usually housed in the wiring closet. The wiring closet may also contain servers that provide resources across the enterprise. In a multifloor design, wiring closets are usually placed one above the other.
Placing the wiring closets in vertical alignment greatly facilitates their connection. Data Center: usually moderately to largely spaced and house all of the necessary networking equipment for the entire enterprise in a central location. As with wiring closets, data centers should be tightly secured and environmentally appropriate for the equipment they house. The data center may contain routers, switches, servers, and even network segment hubs that connect individual devices to their network segment. Rack: Hubs, Servers, Switches, Routers are bolted to them. Packet Errors: Related to Early Collisions and Late Collisions Early Collisions: Collisions in an Ethernet network are to be expected, and the collisions themselves are not a problem. However, when too many collisions occur, say, 5 percent or more of the total packets, then corrective measures are needed.
If this happens too often, the segment network may have to be split. Late Collisions: can be caused by excessive cable lengths. Another potential cause is the use of too many repeaters. Late collisions can result in lost packets that require retransmission by higher-level protocols. Runts: Too small of a packet, may result from a defective NIC. They are also caused when a transmitting device stops transmission in the middle of a packet due to the detection of a collision.
Runts can never be entirely eliminated, because they result from normal collisions, but when the number of runts is greater than the monitored number of collisions, a problem is indicated, may be caused by a defective NIC. Giants: Too large of a packet, and usually caused by a jabbering NIC. Jabbering: NIC is one that is transmitting continuously and incorrectly. Unlike runts, giants are not the result of a normal Ethernet operation, and therefore indicate a definite problem. Whereas a bad NIC is the mostly likely cause of a giant, another hardware device may also be faulty or a cable segment may be defective.
If a NIC or cable segment is found to be the cause of the problem, the best solution is to remove and discard the failing component and replace it with a new one. Broadcast Storm: When the total broadcast traffic reaches or exceeds a rate of 126 packets per second, a broadcast storm results.
The major problem with such a storm is that it is self-sustaining, resulting in a flood of garbage packets that eventually consume all network bandwidth, preventing any other valid communications from occurring. SMDS (Switched Multimegabit Data Services): supports the exchange of data between LANs in different parts of a city or between network segments over a large campus. SMDS is a packet-switched datagram service for high-speed MAN traffic. SMDSIP (Switched Multimegabit Data Services Interface Protocol): provides for three layers of protocols that define user information frame structuring, addressing, error control, and overall transport. SMDSIP Level 1 defines the physical interfaces and the type of trans- mission medium and signaling system used. SMDSIP Level 2 provides an access method, defined in IEEE Project 802.6, that is referred to as a distributed queue dual bus (DQDB).
(It is beyond the scope of this text to go into the details of DQDB; however, it is interesting to note that the access method used is not contention or token passing, but one called distributed queues.) SMDSIP Level 3 accepts user data and adds header and trailer information to it for processing by the SMDS network.