|
|
|
|
|
Building a Network
|
|
|
Introduction This document presents a “standard network model” aimed at a medium sized mainstream college or university. It is intended to identify designs and issues for consideration by local planners when developing a specific network plan. This model does not intend to present adequate detail to serve as a plan itself, but should present a checklist of issues to consider and a set of “rules of thumb” to guide decisions planners must make when developing a thorough campus network design. This model is based on an overall goal of meeting standard network practice today with an expectation of continuing regular upgrades to the network through routine maintenance to maintain full currency with network capability, capacity and function into the indefinite future. This network model will easily support current (2000) network services such as standard Internet services, audio and video streaming, secure ecommerce applications, etc., at an excellent price point. This network model is not intended to support what are currently leading edge applications such as full motion digital video, though to do so would only require adjustments to the network electronics. With routine maintenance based on appropriate life cycle funding for the network electronics, this model will track mainstream network applications into the indefinite future. Target Institutions for the Network Model • ~ 1000 students, 10 buildings, single campus • Can easily be scaled to smaller schools • Can easily be scaled up to 25 buildings or more, or several campuses; more complex installations will need more complex design adjustments. Institutional Coordination is Critical: • Network planners need to coordinate with facilities, data network staff, video staff, and others who will be building or supporting network related facilities or services. • Equally important, all those staffs need to connect with network planners whenever their normal course of work brings them in to contact with network related facilities and services. • This includes not only existing network facilities such as wiring closets, etc., but also new construction, routine or project renovations of spaces and installation of sprinklers, elevators or other apparently unrelated projects which might allow the installation of needed network pathway within or between buildings. • Because network needs are changing quickly, it is important for network engineers to review architectural plans before final approval. Campus Topology The network topologies chosen for this university campus were decided based on the level of sensitivity of the material that would be distributed amongst local users. Also the projected network traffic and geographical location of buildings were major influential factors in the topologies chosen. The sensitive areas of the campus were identified as the “Examinations and Records Department” and the “Accounting Division”. Being precautious and to safeguard from remote hacking of such sensitive material was our priority. Material like Student grades and monetary matters of the university we thought should be connectively and physically isolated from the other sections of the main network of the campus. This is will inevitably prevent students and general staff from remotely accessing these isolated systems from their offices, classrooms or labs. The larger network, which spans over the majority of the physical compound, uses a Hierarchical Star topology. This method allows for more than 12 hubs representing distribution points in each building to be connected to two central hubs, (preferably two switches, for more than 12 distribution points) as the core (located in the Computer Systems Building). The distribution point in each building allows access points for workstations in each needed floor and room, example laboratories, staff rooms, class rooms, lecture theatres. Each room will have its own hub, which allows for multiple computers to be connected in the same room. A redundant Ring topology will connect all the hubs at the distribution layer of the hierarchical structure in the event of failure with the main network connection the network will still be stable. The Computer Systems Building will have several laboratories and Audio Visual rooms, which will also be connected to the hierarchical star network. Backbones will be used to complete connecting the hierarchical star network within the buildings of the campus. Fiber Optic cables will be used to implement the backbone, because the backbone will be located amongst electrical cables used to power the building and this cable is not prone to electromagnetic force interference cause by power lines. Also Fiber Optical cables will allow for the safe distribution on signal with a large enough bandwidth, which may be required in peak usage. The distribution point in each building will be in a secured room being connected by the backbone to each floors and rooms hub. The isolated networks of the Accounting and Examination departments will each use independent Star topologies connected via FDDI cables. This was chosen because of the size of the network, ease of wiring and the fact that a machine can be added or removed from the network without bringing down the entire network. Management of this network is centralized around the hub and switched components. Congestion and network errors can be easily identified by the administrator, which makes troubleshooting quickly and simple. FDDI cables would be used in the Star topology because of the small span of the network and prevent easy undesired tapping of cables to unauthorized users. Fiber Optic Cables will be used to connect the networks, it offers a higher bandwidth, and lower signal disturbances, it also allows higher data rates over longer distances. The 62.5/125 micrometer fiber will be used, it has a minimum bandwidth capacity of 150 MHz per kilometer, and at over 100meters 1.5 Gbps. The low signal noise allows for greater transmission distances and with the use of repeaters a greater distance for transmission can be achieved. The cable is considerable lighter than other mediums and ranges from a maximum 93lbs – 25 pair backbone UTP to 11 lbs two fiber cable. Security is a major concern in this network; it will take extremely expensive equipment and a skilled operator to physically tap the cables. When compare to other methods of transmission, fiber-optic cable is the most secure medium for carrying sensitive data. Projected bandwidth requirement for the future Bandwidth – The range (or width) of frequencies used for transmission of a signal on a network media. This is expressed as Hertz (Hz) as a difference of frequencies. The current design allows for network communication speed averaging 200 MHz. If there is to be an increase in bandwidth it will occur in the backbones of the network. Here a newer 10 gigabit Ethernet wired system will be implemented for instances where mainframes and desktops are connected which are using new graphics or similar software that requires a larger bandwidth than the current system can offer. Scalability Asynchronous Transport Method (ATM) ATM technology is now utilized in many areas of digital communications, and would be suggested in the implementation of a university’s network system. ATM allows for scalability which is needed for a network to grow with the demands that are placed on it. For instance, if there was a pre-existing Network which was a 10 megabit LAN across campus. It can easily be upgraded to a 100 megabit LAN, but that upgrade offer’s little growing room. Upgrading to higher bandwidths requires a change to technology such as Gigabit Ethernet. However, Gigabit Ethernet is currently not available in a form that is useful due to its severe distance limitations. ATM allows the integration of the current 10 megabit Ethernet devices. These devices are grouped into 100 megabit Ethernet switches, using ATM at 155 megabit to back-fed them to the central services area. With ATM the back-fed circuits can be upgraded, in the future, to higher speeds without changing the protocols running on the network. Adding the faster electronics, which do exist today, allows us to scale up the capacity of the network without throwing away the existing equipment. Each department would require different protocols depending on the level of communication. A suggested protocol to be used is RFC1483, which is a protocol for sending IP only traffic across an ATM link, this protocol will give the users of the specified faculty access to the Internet, and other IP based services such as Web and Email, but wont grant access File Server for instance, which is connected via non-IP packets. For the Faculties that need to access both the IP and non-IP services. For these buildings the appropriate services, LECS, LES/BUS pairs, should be assigned to the specific pieces of equipment to run those ATM services. The services are placed on multiple machines, and DLE was used allowing the redundancy and performance improvement of the LANE (LAN emulation) services. The integration of the firewall into the network should be implemented. With the firewall being at the heart of the network, its connectivity to the rest of the network is carefully considered. The placement of the firewall allows it not only to protect the campus from the Internet but also to protect portions of the campus, most notably the central computing facility, from the remainder of campus. One firewall is able to accomplish both tasks. Fault tolerance issues System Fault Tolerance System fault tolerance (SFT). It is the measure of how well your network or stand alone computers can withstand events which cause data loss, lock-ups and system crashes. System Fault Tolerance is all about being prepared. When circumstances arise which can potentially cause a network crash and/or data loss, your system should always be prepared for such circumstances, when hazardous conditions arise, the measure of SFT implementation will determine if your hardware can overcome these conditions or if it will find itself vulnerable. If it is vulnerable, it will determine the degree to which your system will be able to minimize or even prevent hardware destruction and data loss. Two types of SFT are normally considered when implementing a network: energy SFT and storage SFT. This briefly addresses the former Energy SFT Factors Energy SFT considers the quality and quantity of electrical power that supplies your servers, workstations and other network components. It is an extremely important consideration when implementing a network, especially in a universities system. Some of the major causes of power related network failures are: Spikes Spikes are extreme over voltages in the power supply that last for only a small fraction of a second. Causes vary, but the most severe spikes are caused by lightning strikes on or near power lines. Resulting damage can be severe, ranging from data corruption to severely damaged computer components. Surges When a high-magnitude voltage occurs that lasts longer than 1/60 of a second, it is called a surge. Surges occur when equipment connected to the power line has been drawing large amounts of current and is then suddenly shut off. Surges can cause more damage than spikes because of their longer duration. They occur quite frequently in commercial areas. Harmonic Distortion Much of today's office equipment, computers, copiers, fax machines and other equipment containing motorized mechanisms, can cause local deformations in the specific sinusoidal "shape" of the power supplied to computer equipment. This is known as harmonic distortion. It can interfere with network communications and cause transformers to overheat Brownouts Long periods of low voltage power, also known as brownouts, can cause major damage to computer hardware and especially on networks.
|
|
|
|
Still Can't Find What Your Looking For? Then Try a Essay Search!
|
