ITD Information Technologies Department
![]()
Site Networking Guide
(or Ethernet Secrets Revealed)
Written by Jim Levine-Steve Huey-Jason Young , Information Technologies Department
First Written on 11/26/96. Last Updated: 9/11/97 with New Additions under Contents.
Contents
This document is intended to help school administrators, teachers, and technology directors design, plan and install their computer networks or LANs (local area networks) so that they are dynamic and capable of connecting to the District's WAN (wide area network) for not only administrative use but for Internet access as well. Much of the information needed to complete this task is described here in layman's terms and hopefully made easy enough to understand. Each school is different and that is why we strongly recommend that after you read this document and before you charge ahead into implementation, you contact a qualified and experienced network engineer to review your plans and help you with your project. We, at ITD (Information Technologies Department) have successfully helped many schools with their networking projects and are available to help you with yours.
This guide does not provide a blueprint specifically for you, nor would that be practical since there are so many differences among schools. What this document does provide is a basic framework of essential information within which you will be able to construct a plan that meets the unique needs of your site while remaining compatible to those standards and protocols established by the SFUSD.
The focus of this document concentrates mainly on the Ethernet 10BaseT (or unshielded twisted pair, UTP) and fiber cable network topology being implemented throughout the world and supported within the SFUSD. There are schools using coaxial (10Base2) cabling, AppleTalk cabling and others, but we will not be thoroughly discussing them here. Consult with us on these types of network implementations, if your plans must necessarily include them.
Note: Following these guidelines will insure that your network will perform better, have fewer traffic problems, be easier to troubleshoot, easier to maintain, easier to enhance or make changes to and be compatible with the rest of the world.
B: Why install a Computer Network?
Computers are taking over the world! Hospitals are asking patients to swallow capsule computers to transmit bodily functions to nearby monitoring stations. Computers are everywhere! Our students need to be introduced to, if not acquainted with, this technology so that they can be capable of not only surviving but thriving in our advanced world.
These days, when we think of computers, we think of computers connected to a network, usually the Internet. Stand alone computers seem very limited to us now.
Computer networks in schools allow students to communicate (read and write) with not only students and teachers in their school but other people around the world. Computers connected to the school network and the Internet prove to be an incredible source of knowledge for research and other educational tasks. It is educational, inspirational and fun! It is also where many jobs are. Need we say more?
We and much of the world have standardized our network topology on Ethernet. See figure 1 displaying a simple Ethernet network. Ethernet works by connecting computers (printers, file servers, scanners, plotters, or any networkable device called a node) together with Ethernet adapters (or cards or NICs - network interface cards) which have been installed into the computers. These cards have a place (or port) on the back of the card to plug in cables similar to what you use for your telephone. These cables with connectors on the end called RJ45 (or 10baseT) jacks utilize eight wires instead of the telephone's four. The cable used is called category 5 cabling (or CAT-5 for short) taken from the specifications for meeting the requirements of transmitting up to one hundred million bits per second (or 100mbps) across the cable. This is the speed at which your computer, which is connected to the network, sends data through the CAT-5 cable. The cable, from the computer's Ethernet adapter, is lead along its path of no more than 90 meters, through a wall jack, and usually ends up attached to a patch panel in the wiring closet. Patch panels (see patch panels below) are located in wiring closets or designated areas where all the CAT-5 cables from classrooms or offices terminate. Large schools or sites could have more than one wiring closet. The main central area to where cables are lead is called the MDF or Main Distribution Facility. If a school has another wiring closet, it would be called the IDF or Intermediate Distribution Facility.
Figure 1: Simple Ethernet Network
Patch panels, which are devices that help organize cables coming to a central location, have cables from each classroom or area come to them and be punched down or connected to each port on the patch panel and labeled with stickers identifying which port is for which room. Patch cables, which are CAT-5 cables with RJ45 connectors on them, are about three feet long and are used to plug into the specific port on the patch panel and then into Ethernet hubs (or repeaters) which we will discuss next. These patch cables are used to connect computers to hubs. Patch panels are not a necessity but are recommended. They allow computers to be attached (via patch cables) to different hubs possibly on different networks and allow for dynamic and easily changeable network configurations. If patch panels are not used (usually in a non professional network installation), the cable from the computer is lead back to the MDF and plugged directly into the hub. Also, cables coming from the computer's Ethernet adapter, in a professional network installation, will go to a wall jack located near the computer with an RJ45 receptacle in it. The wall jack will then have its cable lead to the patch panel. See figure 1.
Hubs are devices that have many RJ45 receptacles (or ports) on the front of them to which RJ45 cables are plugged. Their function is to take packets of information or computer data received from a computer, through a cable coming into one of its ports, and send out , repeat, or broadcast that data on every other port in the hub. In other words, all ports see the same data. Each packet contains an address which uniquely identifies another computer on the network destined to receive that packet. That packet travels around the network until it arrives at the computer that is supposed to receive the data. Each computer on the network inspects the packet to determine if the packet belongs to itself. If it does, the computer takes the packet in and processes it. If it does not belong, the packet is ignored. Hubs can be connected to other hubs which in turn will broadcast this data through its ports to all of its connected computers. This packet propagation is true as long as a packet of information does not have to travel through more than four hubs to get to its destination. This is called the Rule of Four and must be maintained. Figure 2 displays a valid network with many hubs meeting the Rule of Four. A network can consist of many hubs which come in a variety of sizes (or number of ports). There are 8, 12, 16, 24 and 48 port hubs ranging in price from $100.00 to $700.00.
Hubs can also be stacked, meaning that they can be connected together with special connectors called backplanes so that they appear on the network as only one hub. This is useful when there is a need for many ports and when considering the Rule of Four.
Figure 2: Network with many hubs abiding by the rule of 4.
The figure above shows the number of hops from hub to hub that a packet must travel to get to its destination (say a printer on the other side of the network). A node connected to one of the hubs labeled #1 travels through #2, #3, and the hub #4 with the printer attached to it to request a print function. The X indicates an illegal addition to the network of a hub which would be considered #5. This addition would break the Rule of Four and cause errors, collisions and poor performance on a network. Also the shaded hubs define master hubs that could be located in an MDF and IDF.
You might ask, "What if one or more computers (or network devices) sends a packet out on the cable and to the hub exactly at the same time?". The answer is: "The packets of data collide with each other and cause garbage on the line". This garbage is detected by the Ethernet adapters and they refrain from sending data again until it disappears. Devices on a single network are described to be in a single collision domain. Essentially, the more computers you have on a network, the more collisions. The more collisions… the slower your network throughput. Throughput is based not only on the number of devices on your network but what they are doing. Video, audio, and WWW applications consume bandwidth (a term used to describe amount of throughput… in this case 10 million bits per second). Traffic problems on LANs can be avoided by implementing a well designed network. This includes separating collision domains (discussed later) and maintaining standards.
The following easy to understand excerpt found on the Internet might help to describe how Ethernet works:
Ethernet Operation:
An Ethernet network works as follows:
Access: A node that wants to send a message listens for a signal on the network. If another node is transmitting, the node waits a randomly determined amount of time before trying again to access the network.
Transmission: If there is no activity on the network, the node starts transmitting, and then listens for a collision. A collision occurs if another node also found the network idle and started transmitting at the same time. The two packets collide, and garbled fragments are transmitted across the network.
Collision handling: If there is a collision, the first node to notice sends special jam packets to inform other nodes of the collision. The colliding nodes both retreat and wait a random amount of time before trying again to access the network.
Reception: If there is no collision, the frame is broadcast onto the network. All nodes listen to each packet transmitted. Each node checks the packet's destination address to determine whether the packet was intended for that node. If so, the node processes the packet and takes whatever action is appropriate. If the node is not the recipient for the packet, the node ignores the packet.
Naval Post Graduate School
Advanced Networking
Spring 1996
Note that packets that collide are not lost forever. The NICs are smart enough to detect collisions and will resend that packet again at the next opportunity.
Hubs connect these nodes to the network. ITD (Information Technologies Department) is recommending the use of Allied Telesyn Hubs. They offer great service, maintenance and educational pricing. Each site is also asked to purchase at least one SNMP (Simple Network Management Protocol) managed hub to be located in the MDF. An SNMP hub will allow ITD to diagnose problems on your network from our ITD office to solve problems quickly. It will also allow ITD to log into your hub and gather statistics about your network to determine if enhancements are needed to avoid future problems.
Collision problems, if encountered on a network, can be reduced by using switched hubs or routers. Switched hubs are hubs or repeaters that are intelligent enough to record and remember which devices are attached to each port. A packet's destinations address can be inspected by the switched hub and sent out only to the port that has the computer attached to it. This way each port does not have to see every packet of data on the network. Each port has a dedicated 10 Mbps of bandwidth instead of sharing 10mbps for the entire network across every port. Collisions are avoided using this method. Be careful when considering switch hubs. Some will only be able to remember one device's address per port and have latencies in transmitting. More expensive switched hubs can record many device addresses per port and pass packets rapidly. Consider this when designing a switched hub which will attach to other hubs on you network. Switched hubs range in price from $1,000.00 to $10,000. We recommend installing regular hubs initially. If you then decide that you need something more robust, you can order a switched hub and always take your regular hub and place it in a class room or at the end of a switched hub. Anyway, by the time you need a switched hub, prices will have come down.
Routers are devices that separate networks and collision domains by not allowing traffic to pass through them if that traffic is not destined for another network. Each port on a router is a different network. Each port represents the start of a new network and also the restarting of the count for hubs in the Rule of Four as shown in the figure below. Routers are also used to connect LANs to WANs via data circuits or high speed digital telephone lines. One router must be used on your network if you want to be attached to SFUSD's WAN and the Internet. See figure 1. Routers can filter packets on source and destination addresses, types of packets (i.e.: WWW, FTP, Mail, etc.), and other criteria. Routers also are able to connect networks of different types by converting packets from, for example, Ethernet to Token Ring. See figure 3 for a simple network design using a router.
Figure 3: Simple network design using a router with two separate collision domains or networks and resets the Rule of Four.
ITD has standardized on Cisco Routers for many good reasons and will be reluctant to (if not adamantly decline to) configure or maintain other brands at this time. The Cisco 2514 is the workhorse router recommended for sites. It's able to manage two separate Ethernet networks and connect two remote networks. ITD can furnish you with a sample PO for ordering yours. Currently they are priced at around $2,400.00.
TSU (Terminal Service Unit) is a high speed digital modem that connects your router to a data circuit or telephone line. This enables your network devices to send packets out to our WAN and onto the Internet at fractional to full T1 speeds (1.54Mbs). We have standardized on Adtran TSU-LT's for good reason and can supply you with a sample PO for ordering. Their price is currently around $850.00.
You will remember that the length or distance of CAT-5 cable from computer adapter to hub, through wall jacks and patch panels can be no longer than 90 meters (300 feet). If the distance must be longer than 90 meters, CAT-5 cabling can be substituted with fiber cable or coaxial cable. Fiber cable is used to pass light signals for data instead of electrical pulses using copper in CAT-5 or coax. Fiber cable can be run 2 kilometers while coaxial cabling (10Base2) can be run 185 meters without degradation. Using fiber requires the use of fiber hubs. If fiber hubs are not used, each end of the fiber cable will need to be attached to a transceiver which will convert the light from fiber to electrical signals for CAT-5 copper. These transceivers are priced around $100.00. You will also need transceivers at the computer adapter end as well. Usually, fiber cables are used to connect MDFs to IDFs. Also, fiber cables should be laid and terminated by a licensed installer.
We highly recommend that when installing CAT-5 cable to an area, pull at least two or more cables into that area! Even if your plan is to only connect one network device in that area now, chances are that you will have the need to connect other devices in the future. Remember that you can attach a hub in that area using only one cable if it follows the Rule of Four mentioned above. It is almost the same effort to pull one cable as it is to pull many. Cable is cheap, labor is expensive…. pull more cable!
The above mentioned components are the necessary items needed to install a network which would be capable of communicating with the world. No file server is needed.
D: Defining Your Network Objectives
Now that you are somewhat familiar with Ethernet networking basics, you can define your network objectives. These can be as simple as wanting only one classroom with two computers connected to your network running only Internet applications or a more expansive full school, every classroom, every office wired design. It is best to visualize how your network should look in a comprehensive design. You should have a plan that includes which classrooms will be connected, where those connections within each classroom are to be located, and what functions those computers will perform. It is important to remember (and many schools completely forget) that electrical power is required near computer work areas, MDFs, IDFs, and other areas where network devices will be placed. With these questions answered, you will be able to define a coherent plan for meeting your objectives and a plan that can be reviewed and implemented easily and successfully. You should consider what type of applications or software you want to include in your curriculum. This network vision will help define what type of computers you will need and what amount of bandwidth they will require from your network. This will also help define what type of network components you will need. The funds available are certainly a factor in your decision on how to define your network, but since network components can be added to in piecemeal, additions can be added when money is available. It is important to have a futuristic vision about what you expect your network to look like and what you want it to do in the end.
E: Locating a Central Wiring Closet (MDF)
Identifying the location of your MDF and/or IDF is straightforward. It should be centrally located in the site within 90 meters of each area or classroom you want to connect. It would be helpful to start with a floor plan of your school. Try to locate a classroom office, closet, broom room, or some secluded area at your site which can be dedicated for an MDF. Best places are areas that are secure, well ventilated, with plenty of AC power and have little disturbance. Think how easy or difficult it would be to get cables to other classrooms from this area. This might be accomplished by running cables through a drop down ceiling, raised floor, or down the hall with cables secured to the ceiling. Cables should be located out of harms way. It is recommended that cables be placed in raceways or conduit if money is available.
If your school is large enough to warrant an IDF because you have a large building, outlying wings or unattached bungalows that would require cable lengths to be greater than the 90 meters, then consider locating another wiring closet to be used as an IDF. See figure 4 for a network design using an MDF and IDF. MDF and IDF would then be connected via CAT-5, coax, or fiber cable discussed above.
Figure 4: Network design using an MDF and IDF.
Do a walk-through. Walk the intended wire path and take into account all obstacles the wire will need to travel through or near(concrete walls, asbestos, other wiring, wire mesh, heaters, elevators, etc.). Do not plan to wire or work near anything hazardous or put yourself in dangerous situations! Visualize the wire path from one end to the other. Think about the best ways the wire will be mounted for it's entire length.
Also, the MDF is the place where file servers, network monitoring equipment, central and switched hubs, and other central networking equipment will be located. Make sure that there is enough electricity and space for all these components.
Here are some tips about MDF and IDF equipment rooms. They should be:
Don't hesitate to contact us (ITD), the Educational Technology Team, the Curriculum Office or Facilities Planning for help.
F: Identifying Needed Network Components
Once you have refined the vision of your network, you will need to determine what items you will need to connect your computers together. Those items are: NICs (Network Interface Cards), cabling and connectors, RJ45 wall plate receptacles, patch panels, hubs, patch cables, a router, and a TSU-LT. The discussion about file servers, printers, and computers are beyond the scope of this document and will be left for another time.
Let's review each network item and our recommendations:
DON'T
Recommendations for labeling.
Each individual cable and communications outlet must be labeled. Chaos reigns without labels! Cable labels shall be made with permanent black ink on laminated white labels between 6 and 9 inches from the cable ends. The label information shall consist of the room number of the lab or classroom end, with a dashed suffix distinguishing multiple outlets in the same room. Thus, the first connection in room 245 would be labeled "245-1", the next as "245-2", and so on. The numbering suffix within a room should be initiated with xxxx-1 at the first communications outlet to the left of the entrance door and serialized in a clock-wise manner within the room. If there is only a single connection in a given room, it is to be labeled "xxx-1", in anticipation of future expansion.
In the wiring closets, each punch-down connection is to be labeled on the designation strips and on the cable jacket with an appropriate, permanently attached, label using the same scheme as above.
The RJ-45 end of each cross-connect cable (described later) shall be labeled approximately 6 inches from it's terminal end.
3. RJ45 connectors are the crimp style plastic ends similar to telephone connectors and are used for making your own patch cables. They can be purchased from any computer catalog and come in bags of 10 to 100. The only connectors we do not recommend are AMP connectors. They seem to have developed their own standard and sometimes do not fit into the crimping tool. If you decide to make your own patch cables, you will need a crimp tool which compresses the RJ45 connector onto the CAT-5 cable. These can be purchased for around $40.00 at any cabling store. See figures 5 and 6 for pictures of how CAT-5 cable should be inserted into RJ45 connectors. This method is called 568B and should be used throughout your network. 568B also calls for making sure that the twists in the wire are maintained within a half of an inch of the connector or punch down. This means that you should not untwist the individual wires in the CAT-5 cable too much. The reason for the twist is prevent crosstalk and interference from foreign electrical forces that might corrupt the data. Price: $50.00 for 100 RJ45 connectors. The figure below shows the color scheme for making your own patch cables. Note that the crossover cable or hub-to-hub cable is wired differently and depicted below. It is used to connect one hub to another. Some hubs have a crossover switch at port 1 which can be set instead of making a crossover cable.
Figure 5: Wiring color scheme for crimping RJ45 ends onto CAT-5 cabling (Specification 568B). Note that both ends are truly identical in the first diagram above (RJ-45 Patch Cable). This can be proven by imagining the CAT-5 cable above to be bent in a "U" shape. Then look at the RJ-45 connectors side by side. Think about it!
Figure 6 : Color picture of how cable ends should look when going into an RJ45 connector for a standard patch cable (also following the 568B specifications).
4. RJ45 wall plate receptacles are used in computer workstation areas to connect computers to the network. Computers can come and go in an area but the places for them to be connected will remain for the next user. These wall plates come with you Netday Kit as well and can be purchased easily from computer stores.
5. Patch panels are included in the NetDay Kit and can be purchased easily from cabling stores as well. Punch downs should also adhere to the 568B specifications. Usually all patch panels have a label indicating that it meets 568b specifications. The different colors of the CAT-5 cable are used to color code the patch panel slots and indicate where each wire should be punched down. Follow the color code!
6. Patch cables can be hand made (see our news group article by Steve Huey entitled 'The Unofficial Net Day Wiring Guide' in sfusd.netday.96 at our news server news.sfusd.k12.ca.us) or can be purchased in a variety of lengths. Prices vary by length. For example: a three foot cable might cost about $4.00.
7. Routers, mentioned earlier here, can be ordered directly from Cisco. Contact ITD for a sample PO and contact. Price: currently around $2400.00.
8. A sample PO for TSU-LT's, also mentioned earlier, can be gotten from ITD. Price: $850.00.
Your plan needs to be documented. Put it in writing! Create network diagrams identifying rooms, cabling, MDFs, IDFs, and locations of network components. Once you have a plan and your plan includes being connected now or in the future to the world, you must have it reviewed by ITD. Since you will be connecting to the SFUSD WAN (Wide Area Network), which includes all schools, the District office for administrative needs and email, and the Internet, you will need to be assigned network numbers that must be unique. This plan will also help us determine if your network adheres to the adopted standards of our overall enterprise network plan.
Don't order anything or start work until your plan has been reviewed by ITD.
H: Information Technologies Department (ITD) Services Provided
ITD has been assigned 64,000+ IP addresses by the international committee (the Internic) that governs the networking world. We manage and assign these addresses to sites in blocks of 62 as needed . When the first block of 62 addresses has been used, additional sets of 62 addresses can be assigned. We refer to additional blocks as secondary addresses. By the way, our network mask is 255.255.255.192 which must be used throughout our WAN. I won't explain the number here, but will say that this mask configuration allows us to have 1,022 networks. If you think of our 125 schools, 33 children's centers, 40 remote offices,…etc.. you will begin to see the reason why we might need 1,022 networks especially since each site might have more than one network.
ITD has also received a CALREN grant for a specific amount of free Frame-Relay (high speed digital data circuits) lines used to connect remote sites to our ITD office. These Frame-Relay lines are at synchronous 56kbps speeds and have proven adequate for most schools. These 56kbps lines are bundled by Pacific Bell and delivered to ITD via T1 lines (1.54mbps). We currently have three T1 lines servicing our schools and will be needing more in the near future. When sites show their sincere intentions about establishing their network and provide a documented technology plan to us, we will then order their Frame-Relay line. Pacific Bell is currently taking around 4 weeks to install these lines.
We have asked each site to designate an IRP (Internet Resource Person) to be the first line of contact for help and problem resolution. We provide the IRP with a bundle of information containing a set of installation diskettes (with Internet software) and instructions along with forms which can be copied and distributed to teachers and staff for use with their networks. The software distributed on our diskettes are older versions of software which will enable you to go and get a more recent version if your computer can handle it. Versions of software change frequently and would require us to update our distribution diskettes impossibly often to keep up. We will leave that job to you.
We hope this IRP will help prevent the 6,000 employees from calling our office for software and help, since we do not have the staff for this task. The IRP will also help distribute information and train personnel which in turn will hopefully set up an 'Each One Teach One' philosophy.
ITD has set up a group of Internet service servers which are providing services to all schools, and include:
ITD does NOT try to filter out undesirable, inappropriate or pornographic Internet sites. We initially started to do this but were overwhelmed with the number of sites and their tactics of changing their addresses frequently to avoid filtering. Also, we read about a court decision in which the judge held the provider liable, because the provider attempted to filter access, thereby taking on the responsibility, but did not totally prevent access to all undesirable sites. The judge stated that if the provider had not attempted to filter, the provider would not have been liable. Since ITD does not filter, we have three recommendations:
ITD will help you with your physical network design. It will also help you with identifying network components necessary to implement your design. We will not help you wire your school or install cable to desired networking areas. We will attach, configure and maintain your router and TSU-LT. We will train staff on how to configure network devices with the necessary configuration information like IP addresses, gateways, DNS, network numbers, etc. needed by every device on the network. We will also try to be available for meetings and presentations at schools were decisions need to be made about changes to your network. Feel free to consult with us anytime, but please schedule it with us beforehand, since there are many of you and only a few of us.
Don't hesitate to contact and use the many school district resources available for help in designing or implementing you network. There is the Facilities Planning Department capable of helping you design your network within specifications and SFUSD guidelines. They will be concerned with standards, safety, asbestos, electrical, and more. The Educational Technology Team will be able to help with design along with furniture layout and the many aspects of incorporating technology in the classroom. The Curriculum Office will be able to also help in many ways with training and adding technology in your schools curriculum. Building and Grounds can construct shelving, check electrical connections, and help in a variety of ways. These offices and others are listed in the SFUSD directory and should be contacted for help at any time. We all need to work together!!
The following is a list of helpful Web resources which should be reviewed to add additional help and information to your knowledge base on networks. Most of these were discovered by seraching the Internet with a search tool like Alta Vista on words like: Ethernet, 568b, utp, or other related words.
Among the best, although sometimes wordy, is the thesis written by Bigelow
of the Navy on networks entitled 'Planning and Implementing A WAN for K-12
Schools' at:
http://www.stl.nps.navy.mil/~rjbigelo/thesis/toc.html
See 'NM210: NETWORK MANAGEMENT' an informative and well illustrated
class on networks:
http://www.cit.ac.nz/smac/nm210/networks.htm
Also, 'San Benancio Middle School Consulting Document Table of Contents'
at:
http://www.monterey.k12.ca.us/~dstihler/sbenanci/httoc.htm
'Tutorial on Ethernet' can be found at:
http://142.227.39.2/burridge/tutor/ethernet.htm
'Ethernet -- Access to Ethernet (IEEE 802.3) Information' at:
http://wwwhost.ots.utexas.edu/ethernet/
'K-12 Network Technology Planning Guide' at:
http://goldmine.cde.ca.gov/ftpbranch/retdiv/technology/K-12/NTPG/NTPG.html
We hope this document has been helpful. We plan to continue to update it periodically with the many items we've missed. Feel free to call or mail your recommendations, suggestions, or experiences to us so that we may add them to this document. Thank you.
Since the last update of this document, there have been major enhancements to technology. The three topics included in this addition are:
Fiber (as opposed to copper cabling):
Fiber (or 62.5/125 micron glass multimode optical fiber), as mentioned above, is the mode by which data can be transmitted using light through glass strands, instead of using copper wire and electrical signals. Usually fiber cable is recognized by it's bright orange color. Fiber is about as inexpensive as copper cable to install, less affected by EMF's (electromagnetic forces), can transmit data at greater speeds, and has become more popular to use. Also, as fiber networking equipment has also become more available and very reasonable in price, we are recommending that at least teacher workstation areas be cabled with fiber and also be used to connect wiring closets together.
The length restrictions on fiber (two kilometers maximum) are much longer than copper cable (100 meters maximum) and make it useful in long distance connections. It takes two (2) strands of fiber to make one connection. We recommend that at least twelve (12) strand multimode fiber be installed between wiring closets. This would enable six (6) different network connections to be linked. Installing fiber is easier than installing copper and requires a technician to fasten the ends and test the fiber. This task is not recommended for the layman.
The ends (or connectors) come in a variety of flavors. The most common ends are ST (keyed, bayonet-style) and SC (a push-pull) connectors. Our initial installations of network fiber were specified as having ST connectors since much of the network equipment at that time used these, but now we are specifying SC connectors since the equipment vendors now uses SC as the international standard. We have used fiber patch cords with ST connectors on one end and ST connectors on the other to handle equipment changes. It does work. Terrible how quite often vendors change standards, but... life is change!
The ends of each fiber cable should be terminated into either a fiber patch panel or a fiber wall receptacle. Some installations we've seen just let the fiber end (with the connector) hang from the ceiling or wall. This is not acceptable and should be placed properly in the wall. Also, rounding corners with fiber should not be at an acute 90 degree angle but rather bend in a gradual arc since glass has trouble bending. Always keep fiber receptacles and ends clean and covered since dust will interfere with data/light transmissions.
An excellent source of more information about fiber is located at: http://www.industry.net/c/orgunpro/tia/folswp5
Switches:
Ethernet switched hubs (or switches) are devices similar to Ethernet hubs but have, in addition, intelligence to record and remember which computers are hooked to each port on the switch. By knowing this information, the switch can, after reviewing the destination address in the packet of data, deliver packets of information to the specific port and then to the specific computer. This eliminates extraneous data being transmitted to all other ports as with the non-switched hub. This means that there will be less collisions of data packets and more throughput since 10 or 100 megabits of space is dedicated to each port.
Switches come in a variety of configurations and ports. Some have ports that can transmit 10 or 100 megabits (Mbs) per second to each port (and automatically detect a connection at those speeds). Most have a number of 100Mb ports (either fiber or copper configurable) to be used for connecting to other switches or high bandwidth devices like servers or CD-ROM towers. Some allow the ports to be separated into different LAN's with overlapping ports on each network. Others have available slots that allow cards to be inserted with other types of connection media to be attached at different speeds depending on needs. Mixing and matching is usually needed to accomodate a robust network.
Switches have become very affordable now and can easily replace the non-switched hubs in the wiring closets. We currently are testing and using switches and recommend their use in middle to large schools with collisions/traffic problems. Considerations before purchasing switches are: 1) Make sure that the switch can handle many (like thousands of) addresses (called MAC, not MacIntosh, address) per port. This is needed when connecting the switch with other hubs with multiple devices connected to it. 2) Make sure latency is low when forwarding packets. Switches need time to look at the destination address in the packet to know which port to send it down. This inspection takes time which could slow down your throughput. 3) Other considerations being price, performance, forwarding rate, filtering rate, warranty, support, standards compliance, SNMP management, etc.
We have been testing and using three different vendor's switches in our office and at school sites. They are:
All seem to be working as expected.
We have designed a few networks now and find that each network design requires different network components based on the number of wiring closets, ports, cabling, servers, labs, etc... The switches listed above were selected for applicability as well as price/performance.
Network Design Standard:
Most of the design standards for our networks have been documented in a manual known as the 'SFUSD Information Technology Utility Guidelines' and should be referred to for answers about your network. Lou Webb from the Educational Technology Team has been instrumental in putting this guide together and we highly recommend the read.
In addition, we are now designing school networks with two separate LAN's. One for students and curriculum and the other for administration and teachers. This can easily be accomplished by using both Ethernet ports on the router and by having the teacher's and administrator's cables only plug into hubs connected to the admin LAN ports while the student workstations plug into the hubs for the student LAN. Primarily, this is being done to eliminate admin traffic on the student LAN (and vice versa), secure administration file servers and stations from mischievous students, and to offer different filters on Internet traffic for students and teachers. This means that each wiring closet will have hub ports dedicated to either students or teachers and each connection back to the MDF will be attached to the appropriate router port (Student or Admin). This is why we are recommending multi-strand fiber cables between wiring closets. These strands can be used to connect different networks from the MDF to IDF's and enable us to separate traffic.
Also, as specified in the 'SFUSD Information Technology Utility Guidelines' we are recommending that, when possible, in accordance with length, four Category 5 UTP cables be installed between wiring closets along with the twelve strand fiber. We have found that schools need this capability because much of their existing equipment can only connect via Category 5 UTP cables.
Note: Teachers and administrators should not design their physical networks. This is a job for the professional network designer along with ITD, ETT, Facilities Planning, and site personnel. School site personnel should specify the functional design and make sure that the network designed for their site delivers what was expected. This document is intended to help site personnel understand more about the physical nature of their networks and in turn help them design curriculum based on the technology.
Once again, we hope this document has been helpful and will continue to update it regularly.
At your Service:
Jim Levine, Systems Engineer, Senior.
Steve Huey, Senior Programmer.
Jason Young, MIS Specialist.
San Francisco Unified School District
Information Technologies Department
135 Van Ness Avenue Room 300
San Francisco, CA 94102
Voice: 415-241-6169
Fax: 415-431-8434
![]()
[ SFUSD Home | PR | District Info | School Sites | Programs | Partners | Links | Intranet(SFUSD only) | Feedback]
All contents copyright © 1996, San Francisco Unified
School District. All rights reserved.
Revised 11/20/96
URL:http://www.sfusd.k12.ca.us/resources/ITD/netguide.html