Chapter 8 Network Design
Learning Objectives • Analyze business operations, including organizational structure, communication flow, and mission critical processes • Describe various network design models, including the three-layer, two-layer, and one-layer structures • Consider performance requirements and improvements for given situations • Design a network based on specific business needs
Chapter Overview • In previous chapters, you learned about a wide variety of network technologies. • In this chapter, you will revisit some of these concepts, but with a focus on designing and implementing a network configuration appropriate for a given situation. • Specifically, the steps involved in analyzing the organizational model and designing the network to serve the organization will be discussed. • You will consider the appropriateness of various devices and topologies for given situations. • Performance considerations, including ways to improve performance, will also be discussed.
Design Methodology • The most important goal in designing the network is to serve the organization’s needs. • You should follow a structured approach when designing networks. This will ensure that the network ultimately meets organizational requirements. • Designing a network implementation involves three steps: • Analyze the organization’s requirements • Develop the LAN topology • Configure logical addressing and routing • In the following sections, each of these steps is discussed in turn.
Analyzing Requirements • Analyzing requirements involves getting to know the organization that is requesting your services. • You must learn how the people in the company will be affected by the network changes. • You must also consider the impact of the network on critical business operations. • Areas of emphasis that you must consider when analyzing the business include the following: - Operations - Physical layout - Mission critical systems - Communication flow - Areas of responsibility - Security - Create an organizational map or chart - Future requirements
Developing the LAN Topology • In this course, you have learned about several different topologies that can be used in network configurations. • In large networks, multiple topologies may be used, but the typical configuration is a star or extended star, as shown in Figure 8-1 on the next slide. • In most cases, selecting the correct topology means considering the organizational structure, physical structure, and communication needs of the organization. • In a star topology, deciding where to place connectivity equipment, such as bridges, routers, switches, hubs, and wiring closets, is important.
Developing the LAN Topology Continued • The configuration of the backbone and the layout of the wiring closets is also a significant part of designing the physical topology. • In a later section, you will learn different methods for viewing and implementing topologies. • When network designers finish designing the topology, they document their design plan on a cut sheet, as shown in Figure 8-2 on the next slide. • The cut sheet is a diagram of the floor plan and network illustrating the network connection points, wiring closets, and the location of resources.
Configuring Logical Addressing and Routing • Once you have determined the topology, you can begin to determine the logical addressing scheme. • The logical organization of the network should match the logical breakdown of business departments, such as marketing, accounting, production, etc. • When you implement addressing on a network that uses routing and switching devices, the logical addressing scheme plays a significant role in how the network will perform. • Routers and switches can reduce broadcast traffic, increase bandwidth for clients, and provide connectivity to remote networks.
Configuring Logical Addressing and Routing Continued • The network administrator can use VLANs to reconfigure the logical LAN topology without making changes to the physical topology. • A VLAN can reorganize broadcast domains, which makes it possible for the logical topology to change with the organization. • Consider these questions before you implement addressing: • How many people will the organization add or lose in coming years? • Will the organization add new offices? If so, how many and where? • What services will be required by users in each location? • A limited implementation or addressing scheme could cause several problems in the future.
Understanding The Organization • The network should serve the needs of the organization, not the other way around. • It is important to understand the organization before designing, implementing, or upgrading the organization’s network. • If there is an existing network and network administration team, they may already have some of the necessary information. • Like many complicated tasks, ensuring that a new network serves the needs of the organization can be more involved than originally anticipated. • The following sections discuss the many items a savvy network administrator should consider when designing a network.
Organizational Structure • Many organizations document their structure in an organizational chart. • You can use the chart during your investigation of the physical layout of the company and the flow of communications. • Figure 8-3 on the next slide shows a sample organizational chart. • The purpose of the organizational chart is to illustrate lines of authority and responsibility, not communications flow. • Unfortunately, lateral communications, which are conducted between peers, are rarely depicted on organizational charts.
Communications Flow • Typically the lines of communication evolve from necessity and convenience and are rarely formally documented. • Understanding where these lines of communication exist are critical to your success in designing the network. • There are two difficult tasks in analyzing the communications flow in the organization. • The first is finding and defining all the flows of communication. • The second is determining how the network can change those flows of communications. • For example, if department heads gather each Monday morning for a meeting, video teleconferencing over the network may provide a more efficient method for this communication.
Communications Flow Continued • While communication flows can be changed, they can also be eliminated altogether. • For example, if the customer service department routinely calls the marketing department to obtain customer account information, a shared database between these departments could eliminate the need for such communication. • Preferences are important. Some companies prefer to have face to face meetings on a periodic basis. Others have people in remote locations with which some organizational members never meet face to face. • The network should serve the communication and information needs of the organization, not the other way around.
Physical Layout • The organization’s physical layout is very important when designing the network structure. • If a company occupies several floors on a certain building, you must consider the implications of forming network connections between those locations. In addition, if the company has, remote locations, WAN links must also be considered. • UTP cabling is typically not the best option for elevator shafts because of the sources of electromagnetic interference (EMI) in the shaft. These sources include the electrical wiring, the motor to control the elevator, and the braking systems. • Typically, network administrators use shielded cable or fiber optic cable in such environments.
Physical Layout Continued • Once you have made these assessments, you will be able to narrow down some of the connectivity options. • For example, if the organization has a couple of offices in a nearby building that is over 500 meters away, as shown in Figure 8-5 on the next slide, connecting remote offices to the main location via UTP wouldn’t be a viable option. • Once you know the various ways in which the network can and cannot be connected, you can determine the topology and media necessary to configure a viable network layout.
Vital Operations • Every organization has a set of operations that is vital to its operation. • Typically, these operations enable the organization to produce its goods or provide its services; they are mission critical. • When designing a network implementation, you should give priority to mission critical operations and the systems that support them. • Investigating the communication flow of the organization will often reveal which systems are vital to the organization.
Availability and Recoverability • Once you have determined which systems are critical to the organization, you should evaluate the importance of each system or group of systems and consider the various methods for protecting the system and making it available. • For instance, in certain organizations, the service goal is 100 percent availability, and you may have to configure redundant systems in order to provide that level of service. • You have various options for creating reliable network services and protecting your data: • Clustering • RAID • Backups
Availability and Recoverability Continued • You should also consider if the servers will require multiple communication paths in the event that a network connection fails. • For example, it may be wise to place two network cards in a system that is utilized by all network users. • This would ensure that there are multiple communication routes available in the event that a single communication route fails, as shown in Figure 8-6 on the next slide. • Companies often implement multiple backup techniques, such as RAID and tape backup.
Routine Maintenance • You must also consider the maintenance tasks that will have to be performed on the network and how those tasks affect the network users. • For example, file backups are a routine task that occurs in most organizations. • You must determine whether the backup should be done centrally over the network or locally at each office or department and at what time. • Some organizations have large databases that must be routinely transferred from one location to another which affects bandwidth. Bandwidth concerns will be discussed in greater detail later in this chapter.
Future Operations • If you are performing an upgrade to an existing network and do so without regard to the future, the planned installation or upgrade can quickly become outdated. • If your design fails to address the growth needs of the network, your work may be considered a complete failure. • Asking the right questions regarding growth will keep you from creating a brilliant, yet inflexible, network that must be done all over again in three months. • You must build a system that is flexible enough to handle the demands of the present and the expected demands of the future.
Reviewing the Existing Network • Upgrading an existing network has its advantages. • For instance, you can benefit from eliciting complaints about the system from network users. • If there are critical systems in use on the network, you can analyze their efficiency, note their effect on the network performance, and check for potential availability and security problems. • Unfortunately, an existing network can also be a hindrance. • One major problem may be that network users have already adjusted to the inefficiencies of the existing network, such as high latency. Even an efficiency change may be unpopular.
Design Models • There are two basic design strategies that are typically followed: mesh design and hierarchical design. • Mesh designs are less structured than hierarchical designs. In a mesh design, there is typically no clear definition where certain network functions are performed. • Routers in a mesh design act as peer devices and perform essentially the same functions. • As shown in Figure 8-7 on the next slide, the mesh is a flat structure in which expansion of the network is done laterally.
Design Models Continued • Hierarchical designs are more structured and defined than mesh designs. • Compared to a mesh design, a hierarchical design: • Is easier to manage • Is easier to troubleshoot • Has improved scalability • Allows easier analysis • In the following sections you will learn more about three hierarchical network models: the three-layer network model, the two-layer network model, and the one-layer network model.
Three-Layer Network Model • The three-layer network model is the most complex of the three models. It consists of a core layer, a distribution layer, and an access layer. • The following list describes each in turn: • Core layer: Provides WAN connectivity between sites located in different geographic areas • Distribution layer: Used to interconnect buildings with separate LANs on a campus network • Access layer: Identifies a LAN or a group of LANs that provides users with access to network services • Figure 8-8 on the next slide illustrates how each layer would be categorized in a large network environment.
Three-Layer Network Model Continued • The core layer connects networks in different cities, the distribution layer connects several LANs within the same part of the city, and the access layer is where users gain access to the greater network. • Each layer is separated into its own broadcast domain because the layers are separated by routers. • Notice that within the access layer, there are several broadcast domains, each providing users access to the network. Therefore, the separation between the layers in the three-layer network model is mostly due to function. • There is also a separation of broadcast domains between the layers because of the connectivity equipment used.
Core Layer • The core layer provides WAN connections between the various locations of the network. Since the core layer mostly consists of point-to-point WAN connections, hosts are not typically part of the core layer. • Organizations usually lease the connections used for the core layer from a telecommunications company. • Organizations use the core layer to provide a fast connection path between remote sites. • Efficient use of the bandwidth is a concern for those negotiating, designing, and administering the core layer connections. • Often, network administrators establish multiple connections at the core layer.
Distribution Layer • The distribution layer consists of all the equipment necessary to make the connection between different LANs in a single geographic area or campus. • This includes the backbone connection and all the routers used to form the connections between each LAN. • At this level of the network, network administrators usually implement policy based connectivity. • This means that the network administrator determines the type of traffic to allow on the backbone. • The network administrator can use routers to filter the traffic from the incoming WAN connection and between the various LAN connections.
Distribution Layer Continued • The administrator can also control the LAN interconnections by establishing path metrics between the LANs. This allows the administrator to predict and control the path a network packet should traverse between two points. • By doing this, the network administrator may reserve certain network communication paths for certain types of traffic. • For example, the administrator could filter all traffic out of a specific path except e-mail communications. • Since e-mail is transferred over TCP port 25, the administrator could create an access list that allowed only TCP port 25 traffic. • The access list filter command would look like this: • ip access-list 101 tcp permit any any eq 25.
Distribution Layer Continued • This would mean that no matter how congested the rest of the network became, e-mail communications would have a dedicated path across the wire. • You should not place any end stations at the distribution layer because those stations would have access to all LAN traffic, thereby creating a security threat. • You should not place a server at the distribution layer because traffic would be originating and terminating at the distribution layer. This compromises the separation of the layers. • For example, if you place a Web server at the distribution layer, users will only have access to it if you allow Web traffic through the routers that separate the distribution and access layers. This means that you can no longer filter Web traffic between LANs.
Access Layer • In the three-layer model, the end systems should be located in the access layer. • The access layer provides a logical grouping of users by function. This results in a logical segmentation of the network. • This segmentation is typically based on boundaries defined by the organization. For example, the marketing, accounting, and human resource departments in a company are typically segmented from one another. • While each would be part of the access layer, connectivity devices such as bridges, switches, or routers would typically separate their broadcast domains. • The main goal at this layer is to isolate the broadcast traffic between the individual workgroups, segments, or LANs.
Other Network Models • The three-layer network model works well for large network environments. • Most network administrators find that the two-layer or one-layer model works best for smaller network environments. • These two models are discussed in the following sections.
Two-layer Network Model • The difference between the two-layer network model and the three-layer network model is that the distribution layer or campus backbone is not defined or implemented in the former. • Network administrators use WAN connections to interconnect separate LANs, and VLANs to define separate logical networks, as shown in Figure 8-9 on the next slide. • Notice that each site is still separated by layer 3 routers, yet the sites may be further subdivided by switching equipment and, as previously mentioned, by VLANs. • The structure is still in place so that the network administrator can add a distribution layer and implement a three-layer model at a later date.
One-layer Network Model • Smaller networks can employ a one-layer network model design strategy. • The one-layer network model has less need for routing and layer 3 separation than the two-layer or three-layer models. • Typically, the one-layer network model includes a LAN with a few remote sites. • In the one-layer network model, servers may be distributed across the LAN or placed in one central location. • Figure 8-10 on the next slide illustrates a one-layer network configuration.
One-layer Network Model Continued • Most of the traffic in the one-layer network will be concentrated on the LAN. • The WAN should only have a light traffic load because the remote sites will not be generating much traffic. • The network users gain access to the rest of the network via routers that may connect other segments of the LAN or parts of the WAN. • Network administrators can change the one-layer design model into a two- or three-layer structure as the network grows. • The main difference between the hierarchical models discussed is the number of routers deployed in each configuration.
One-layer Network Model Continued • In the three-layer model, there are three levels of routers: one at the core layer for the WAN, one at the distribution layer for the campus backbone, and one at each site in the access layer. • In the two-layer network, two levels of routers divide the structure. • The one-layer model only has one-layer of routers, all of which operate at a peer level. • The main difference between the one-layer design model and the mesh design is the definition of structure. • With a mesh network, the chances of having to restructure the entire network configuration as the network increases in size are greater.
Network Media • Today, networks use a variety of connectivity media in addition to network cabling. For example, infrared networks, electrical circuit networks, and satellite transmissions are just some of the ways in which network devices are being connected. • The most frequently defined and popular networks are still installed with copper and/or fiber optic cable. • We begin the discussion with issues that you should address when deciding on the type of cable to use: Distance limitations Expense Potential sources of interference Resistance to interference Routing Security Installation Existing wiring
Cabling Standards • As you have learned, EIA/TIA-568 and EIA/TIA-569 standards describe horizontal cabling specifications. • Horizontal cabling is the twisted-pair or fiber network media that connects workstations and wiring closets. • The specification covers the outlets near the workstation, mechanical terminations in the wiring closets, and all the cable that runs along the horizontal pathway between the wiring closet and the workstation. • The standards also specify the names given to cabling and devices on the network, as shown in Figure 8-11 on the next slide.
Cabling Standards Continued • The figure shows the following items: Workstation area cable Telecommunications connector Horizontal cable Horizontal cross connect patch cord • EIA/TIA-568B specifies that the maximum distance for a UTP horizontal cable runs is 90 meters (295 feet). • The patch cords located at any horizontal cross connection cannot exceed six meters (20 feet). Also, patch cords used to connect workstations in the work area can be up to three meters (9.8 feet). • The total length of patch cords and cross connect jumpers used in the horizontal cabling should not exceed 10 meters (33 feet).
Cabling Standards Continued • Based on the previous specifications, the network designer should not make the total length of any segment exceed the 100-meter limitation for UTP. • In addition to UTP, the following cable types may be used for horizontal pathways: • Shielded twisted pair (STP): Two-pair of 150 ohm cabling • Fiber optic cabling: A two-fiber 62.5/125 µ optical fiber cable • Coaxial cable is also part of the specification but is not recommended and will soon be dropped from the standard.