Note: Does not include latest updates including category 6 cables, Gig-E, 10Gig-E
 

1. GENERAL 1.1. Introduction A structured cabling system will generally be provided as part of a computer networking strategy. Where new buildings are concerned, the scope of the structured cabling system may be broadened to encompass all communications services, including voice, video and data. In existing buildings, it may be desirable to migrate existing voice services to structured cabling to facilitate an "Integrated Voice and Data Cabling" strategy. These notes outline planning methodology applicable to a multi-building environment, with an emphasis on computer networking. For simplicity, only Cat 5 Unshielded Twisted Pair (UTP) solutions for workstation outlets are considered. This may not be appropriate with some equipment types or in some electromagnetic environments. Proper engineering advice should be sought when in doubt. We can provide advice and design based on faxed, mailed or E-mailed plans and text.   Specification text and CAD work including digitization will generally be delivered in electronic file format as E-mail attachments for remote printing, or by courier or fax as completed work. Rates are by negotiation, but will generally be under $ 80.00 per hour. Send requests for fee proposals including project brief by E-mail to info@technixs.com 1.2. Standards An understanding of the applicable standards is essential, and people wishing to embark on complete design and installation themselves should obtain a copy of the EIA/TIA 568 standard. The standard details cable types and performance, connector standards and pin outs, general structured cabling architecture and distance limits, and link performance. Categories and link classes are defined, the highest at the moment being Category 5 and Class D with performance defined up to a frequency of 100 MHz. These standards are completely unrelated to data networking systems and protocols. Today data networking standards are commonly being developed on the assumption that a structured cabling scheme will be in place to transport the data. Cable pathways including underground routes, conduits and tray ways, access flooring and equipment closets are covered in EIA/TIA 569. This standard is a guideline, not a prescription for a solution. Various web sites provide overviews of various aspects of structured cabling systems. A suggested starting point is Siemon's overview at http://www.siemon.com/568over.html. 2. PLANNING A COMPUTER NETWORK 2.1. Standard Computing Environment and Network Type It is assumed that one of the objectives of the Network implementation is to support an Ethernet standard networking environment, which is based on International Standard IEEE 802.3 ("Ethernet") networking products and protocols. For more information on all aspects of Ethernet from first versions through to Gigabit Ethernet, visit the University of Texas Ethernet site at http://wwwhost.ots.utexas.edu/ethernet/. 2.2. Objectives The first step is to establish the aims of a local area network implementation. These might include:- Implementation of administrative and financial database Staff access to company records Automation of letter, report or specification writing E-mail for staff Staff scheduling General information automation (including library, plans, graphics and images) Learning or training aids (interactive software) Computer skills training rooms (word processing, publishing, CAD, spreadsheets, databases) Printer sharing File transfer Internet access (graphical, text, news) Access to centralized information sources (e.g. CD-ROM stacks) Automate software updates Centralize application software 2.3. Choice of Software and Hardware Before considering network requirements, the machines and software which are to be networked now or in the future must be identified. The purpose of this step is to consider what existing hardware and software is available in order to make an economic analysis as well as keeping the scope of the project reasonable. Identify which software applications the network operating system and hardware must support Exclude software or machines which will be discarded for other reasons from further networking considerations. After answering the following questions, it should be possible to identify which PC's will initially be networked, and what existing "legacy" networks should be supported and grafted to the new network. a) Which software packages are proposed to implement the target applications? b) What hardware platform (type, size and speed of PC) will be required to run the software? c) Can existing computers be used, or will they require replacement? d) Can existing computers be upgraded (higher speed CPU, upgrade OS, add NIC to support networking, etc)? e) If existing computers require replacement, should they be redeployed to less demanding tasks? f) To what extent will expenditure on replacement PC's and software reduce the available budget for networking? All the above options have some monetary, compatibility and user interface implications, all approaches may be quite substantial compromises. However, there is no point on spending a good deal of money on systems to incorporate Apple computers into an otherwise PC network if the end result does not achieve network objectives. 2.4. Mapping the Network 2.4.1 Cabling Scheme The main elements of the standard structured cabling schemes are:- Fixed building cable comprising 4 pr single strand Unshielded Twisted Pair Cable to Category 5 specifications wired from a wall outlet to a patch panel mounted in a hub closet. Termination at either end in a standard sequence (EIA 568) and pin-out on 8-pin "modular" (RJ45) connectors. Ethernet patch cords comprising 4 pair multi-strand Unshielded Twisted Pair cable straight through connected to RJ45 plugs at either end. Used to connect patch panel to data hub, and wall outlet to PC's NIC. Depending on the requirements, Singlemode (8/125 um) or Multimode (62.5/125 um) fiber optic cable may be used and typically will use snap-in SC duplex connectors radiating out from a central location or campus hub to all other hub closets in a logical star fashion for distances up to several km. All fixed cabling terminated on patch panels, and connected to equipment using patch cords.  Proper labeling at the time of installation is a must! Most data networking equipment being sold or designed today is based on the assumption that the above cabling scheme will be available to interconnect the equipment. The standard is quite emphatic in its preclusion of fixed cabling being connected directly to hubs or to PC's (eliminating patch panels and wall outlets). Difficulties are experienced with outlet identification, reliability of plug termination and breaking of single stranded cables over time, with such arrangements. Where the cabling system is connected to equipment linked to other premises (e.g. offices or servers in other cities) via carriers (e.g. the ILEC, CLECs or IXCs), the cabling scheme itself must comply with Telco industry standards. 2.4.2 Node locations Having selected which PC's, printers and dumb terminals are to be connected to the network initially or in the future, all such network "nodes" must be located on a plan of the campus which is approximately to scale. A "saturated" structured cabling scheme would allow for voice and data outlets at every potential work desk or computer location throughout the site, wired back to respective nodes. 2.4.3 Locating Hubs Each of the above nodes will be connected to an interconnection box called a "hub". The hub contains repeaters, patch panel for outlet cables, patching for coaxial or fiber backbone cable, and may also contain switches, bridges and routers. One "campus" hub will be the network's focal point for the campus and might take up as little space as a large filing cabinet up to a fully fledged computer room, whilst the other "workgroup" hubs might range from one or more equipment racks down to a collection of wall mounted boxes and equipment occupying a wall space of about 10 sq. ft., and about 15 in. deep. Ethernet switches have dropped in price and improved in functionality greatly.   These offer considerable advantages over simple Ethernet hubs for medium to large networks.  Big switches (able to connect many PC's) are generally cheaper per port and more feature rich than smaller ones, so as a rule of thumb, it is desirable to have at least 24 to 36 nodes connected to an Ethernet switch. It is desirable to have hubs in rooms or closets which are secure from unauthorized access. It is also desirable to have such hubs at least 3 ft. away from any other electrical equipment. Select one hub location as the major network hub. This should be located at a point within easy reach of the person likely to be responsible for network administration, but it should also be readily accessible for adding future backbone cables, and preferably towards the center of the campus or building(s). On large sites it is generally the computer room. Hubs may serve more than one building, providing a viable cable path can be found between buildings and both buildings share the same electrical earth and distribution. Possibilities include:- Install underground in existing or new conduits. New conduit usually required, and most easily installed if route is grass or garden. Special underground cat 5 cable must be used to guard against moisture damage. Run in conduit or clip on cover rectangular duct under covered walkways Overhead catenaries between buildings (probably the only practical approach with transportable buildings). 2.4.4 Selecting Backbone Routes Having located hubs, a means of connecting all the workgroup hubs back to the major network hub must be found. The structured cabling system approach is to provide separate fiber optic cable pairs (one for transmit and one for receive) radiating out from the campus hub to each workgroup hub. Where voice traffic (telephone, ISDN, etc) is also reticulated through the scheme, multi-pair cable is also provided for this purpose from a PBX frame/patch panel or MDF to a patch panel or "pair management frame" at each hub location. Usually one or more spare fiber pairs would be provided to each hub location in case segmentation of the network occurs in the future, or to make direct connections to remote file servers. If hubs are in the same building, it might be possible to link them using Cat 5 cable providing equipment at either end is suited to it and there are no lightning protection, surge, or grounding issues which could cause problems (generally more applicable to multi-story buildings). The way the fiber optic cable is laid depends on the topology of the campus, and the staging of network implementation. Possibilities include: Cluster style campus, individual cables radiating out from campus hub like spokes on a wheel. Ring style campus:- Cables radiate around the ring, possibly all being laid in the same conduit or duct until they exit to a workgroup. A multi-fiber cable may be laid along part of the route to a workgroup patch panel, which is then patched to smaller individual cables continuing on to other hubs. Straight line type campus:- As for ring, except cables are laid in a straight line along the campus. 2.4.5 Linking Workgroups at the Campus Hub Although fiber will handle very high data speeds, in reality, the speed of data on a fiber is governed by the available data interface to it. The standard interfaces are 10 or 100 Mbit/sec just like twisted pair or coax. Higher speeds can only be achieved by purchasing ATM (155 Mbit/sec) or the newer Gigabit Ethernet (1000 Mbit/sec) network equipment, which are generally only economic if the network has hundreds of workstations. All network traffic between workgroup hubs passes through the campus hub. A means of connecting all the fibers radiating out to the hubs is required. The connection can be made in a number of ways:- Multi-port fiber optic repeater:- Broadcasts data from any connected workgroup/fiber to all others. Unless bridges are located elsewhere, all parts of the network will carry the entire campus traffic. Multi-port switch or bridge:- Only allows traffic from workgroup connected to any particular port intended for other workgroups to pass between ports (stops local workgroup traffic from slowing down the campus network). Some switches/bridges can filter traffic to prevent nodes from accessing selected other nodes or have VLAN capability (distinct virtual LANs). At a switch or bridge, traffic is regenerated, so in terms of network rules, the bridge is the start of a new network. The market is predominantly for switching these days. Switches are most effective when used with centralized file servers connected directly to switch ports (e.g. all file servers in one room or area) and on medium to large networks (over 25 PC's). Multi-port Router:- Decodes source and destination node addresses of data packets incoming, and forwards them to intended work-group and node. Can have very complex restrictions on the type of traffic which can pass through, and who can talk to who. As for switch, traffic is regenerated, but with some time delay. Can also change network data speed for connection to lower speed outside links. Usage as for switch/bridge. The minimal installation for any site with more than two hubs is a multiport repeater. Two hubs can be linked by cascading one hub off a single repeater port on the other hub. 2.5. Implementing the Network 2.5.1 Network Interface Cards (NIC) With the exception of already existing networks, all PC's and network printers to be connected to the network must have a 10BaseT or 100BaseT network card fitted. Some recent model PC's and Macs may have NICs already built on to their motherboard. Standard Ethernet network cards provide a 10/100BaseT connection to the PC. There are a number of types of network cards for PC's, which mostly vary with the type of internal PC bus. The following bus types will typically be found: ISA 16-bit bus EISA 32-bit bus (obsolete) PCI 32-bit bus A list of all equipment to be fitted with NICs must be created, together with type of NIC proposed. It must be decided who should be responsible for installing and setup (company employee or a contractor). 2.5.2 Checking Proposed Approach After creating a draft plan of action, it will be desirable to verify that the proposed network architecture is practical, and to confirm that important issues haven't been overlooked. Options for checking proposals include:- Engage independent consulting engineer to do a review Engage software consultant or vendor(s) to review hardware approach. Discuss with potential equipment suppliers and cabling contractors Discuss with IT Branch of organization Discuss with comparable businesses who have gone through a similar implementation Engaging consultants will involve expenditure, but may speed the deployment process up by allowing key issues to be quickly identified, and questions to be directly and independently answered. Where the proposed expenditure exceeds about $20,000, a consultant review is strongly recommended. Consideration might be given to digitizing campus plans from existing fragmented paper copies. This can then be used as a master plan and kept up to date as the network develops, as well as a basis for specifications. A building outline site plan for a large campus would cost about $350 to produce, while a plan including all room divisions would range from $450 to about $5,000 depending on campus size. 2.5.4 Specifying Cabling A specification must be produced for the cabling work required. The specification will generally include: A site plan showing backbone cable routes and hubs, pits and existing conduit locations, if to be reused, (also size and type of conduits to be used in new installation if possible). Underground routes should be clearly identified. Scale floor plans showing outlet locations (use varying symbols to distinguish single sockets from multiple sockets) Schematic diagram (a graphical summary of the proposed network showing number of cables or cable pairs over each route, and cable type; also showing how existing cabling is to be joined to new, if applicable). Schedule of equipment to be provided (fly leads, patch cords, hubs, equipment racks) Text outlining standards to be met (e.g. EIA 568C Category 5), general workmanship and materials to be used (e.g. style of outlets), any special details not easily shown in graphical form. Specify underground rated cables capable of resisting moisture ingress when lying in water for underground routes. Documentation to be provided at the end of the job (test results, outlet numbering plan, actual cable types used marked up on plans) Sample drawings are shown in the case studies, and a generic specification is given in section 3.3.1. Depending on skills available within the business concerned, assistance or review by consulting engineers may be desirable. 2.5.5 Linking Buildings If links between buildings are required as part of the project, a number of decisions need to be made to select and specify the appropriate cabling. These include: Will link be run overhead? If so, will it be enclosed in conduit on a catenary, or will it be outdoor rated cable with integral catenary? Will link be run underground? If so, it will need to be jelly filled moisture barrier cable, usually loose tube construction. Is an existing underground route available? If not, planning should allow for other compatible services over the same route, such as security, CATV, PA, voice, control and, of course, existing utilities (e.g. telephone, gas, electric, water). Will the route be subject to rodent attack? If so, armored cable or enclosure in conduit will be required. Will the cable be required to carry broadband TV or gigabit Ethernet in the future? If so, it should contain some single mode fiber pairs. Is the route length (including intermediate patches) back to the campus hub over 1,000 ft.? If so, consider adding some single mode fiber pairs. 2.5.6 Selecting and Purchasing Hub Equipment While cabling is being installed, hub equipment can be purchased. This equipment will interface to all the cabling provided, and transfer data traffic between workstations, file servers, and internet links. Draw a schematic diagram of the network proposed showing work-groups, file servers, printers, hubs and backbone routes, in a way which illustrates the functional and physical relationships of the workgroups. Create a list of what applications the network is to run, what the preferred NOS selected will be (e.g. WinNT or Win2000, Novell Netware, etc), and what the connectivity objectives of the network are. Vendors can then be approached to provide suitable equipment to utilize the cabling infrastructure provided, and meet the stated objectives. Include installation and commissioning if necessary. Hub equipment required will be as follows:- Hub backplanes or chassis: required at each hub location where plug-in style hubs are to be used. Provides common power supply and interconnection medium. Workgroup Concentrators: Alternative to hub chassis. Generally used with stackable hub systems, and comprises repeater, fiber optic backbone interface, power supply, and network management interface in one box. Miniature (8-port or 16 port) local repeaters: Used to connect a cluster of PC's in a room to a single hub closet repeater port, or where 15 or fewer PC's are likely to be connected at a hub location. Use outboard fiber optic to 10BaseT or AUI converter to connect to fiber backbone if necessary. Fiber Optic Media Interfaces: Required to connect hub to fiber. Basically a single port repeater or plug in circuit adapter for a hub. Fiber Optic Multi-Port Repeater or switch: Used at campus hub to interconnect fibers. Switch:- Used to isolate backbone workgroups and servers to allow concurrent sessions within and between segments Higher speed networks: Should a need for higher speed be confirmed on the backbone, partial or total implementation of Gigabit Ethernet might be considered. See http://wwwhost.ots.utexas.edu/ethernet/descript-100quickref.html for a more detailed consideration of design rules. This networking methodology imposes a number of equipment architectural constraints which mean a 100baseT/F network can't necessarily be scaled up directly. The number of different equipment types used should also be minimized, so that equipment is easily interchangeable around the campus, and equipment replaced in one area can be used to boost capacity elsewhere. If spares are to be held on site, equipment rationalization will minimize the inventory. 3. SYSTEM ADMINISTRATION 3.1. Documentation It is essential that documentation be kept for a variety of reasons. Good documentation will: Enable a completely unfamiliar person to quickly grasp the network topology. This is most important in an environment where frequent staff changes can be expected. Help the network to grow in a planned and well structured manner, allowing the best use of available budget. Assist others involved with your network such as cable installers, network trouble-shooters and consultants. By enabling them to see exactly what is in place, they will save time and understand exactly what is required, which will result in cost savings. Provide a valuable tool for locating faults when things go wrong. Assist recovery of insurance in the event of fire or theft. 3.1.1. Details to Record The documentation need not be onerous or overly formal, but at a minimum the following details should be kept together, preferably in a loose leaf binder. Drawings showing the location of cable runs, (preferably with individual cable lengths marked for any legacy thin Ethernet runs). Size and occupancy of any existing cable ducts or conduits which are available for use when expanding the network. Details of the capacity and type of cabling and connectors used. Details of type of connectors or sockets at either end of each cable (blanket statements adequate). Records of the designation and location of outlets. Any certification carried out on the cabling (e.g. Cat5 test results). Diagrams showing the relationships of various workstations, file servers, printers and other devices on the network to one another. Diagrams showing how any hubs, bridges, switches or routers are used to connect the network (may be shown on a network management system). Records of software licenses and versions of software installed. Configurations of NICs in workstations. Records of computer numbers and types, serial numbers, disk drives installed, printers, modems, hubs etc. 3.1.2. Patching and Jumper Records Patching records should be retained at each patch panel or cross connect. These should identify relationship between hub and switch ports, patch panel ports, and end user devices. Spreadsheets can be used to automate this process. If available, databases and cable management systems may be used to provide added sophistication. 3.2. System Administration In most instances networking is not simply a matter of connecting a group of computers together at a hub. The network requires special software called a Network Operating System to allow communication between the various devices. The NOS software does not simply run itself either. It requires a person called the System Administrator to perform administrative functions using the NOS software to do such things as make backups of files, keep the traffic flowing smoothly, and ensure various users have authorization and access to communicate with printers, the Internet, and other computers. When the computer system `crashes' the System Administrator will bring it back to life. When new users or equipment are admitted to the network the System Administrator must update records in the NOS to allow the new users or equipment to be used. The amount of time and skill required of the System Administrator depends on the size of the network. For a network of 10 computers perhaps only an hour per week would be required on average. For a network of 100 computers, more like 10 hours per week would be required on average. More complex networks involving bridges, routers or Internet servers require more time and skill than simple networks. 3.3. Maintenance and Repair Inevitably some equipment will fail or need upgrading. A large network of more than 100 computers may require a technician for 20 hours per week. For a small network of 10 computers it may be more cost effective to have a competent member of staff perform the necessary upgrades, or take the equipment to a repair agent. It is worthwhile considering how these issues will be addressed, and perhaps whether a service contract is required with a computer supplier or agent.