A Guide to Networking a K-12 School District
by
Bradley H. Lamont
Implementing a network plan
Once a district has developed a network plan for each of the buildings it will network, the next step is to replace the generic parts of the plan with actual equipment and wiring as shown in Figure 3-1. Following that is the installation of the equipment and wiring in a building. When this is completed, the network is ready for use. This chapter focuses on the different options available for implementing a network plan. It discusses guidelines for the installation first. Then it presents the different equipment used in a LAN. After that, it presents a similar discussion about WAN equipment. The chapter concludes with an examination of the role software plays on the network.
3.1 General guidelines
Once the network plan has been developed for a building, it is time to install the network. Before starting the installation, time spent evaluating how the installation will proceed can save significant amounts of time and headaches later.
3.1.1 Suggestions
The first step should be to obtain accurate plans for the building. Many districts do not have plans for some of their buildings, so the designers should use whatever maps are available. In either case, be sure to verify the distances before installing the wiring since the maps may be inaccurate or out of date. This can be done using a distance wheel, often available from the athletic department of a school. In areas where the designer can not take accurate measurements, estimate the distances but try to be on the conservative side. Be sure to never estimate wiring distances to the exact foot since errors will always appear and measuring too accurately may not leave enough room to work around the problems. Remember that besides the horizontal distances, the wiring will also need to run up and down walls. Always leave 10 to 15 percent of the maximum distance available unused to allow for the distance needed to connect a machine to the wall and to connect equipment at the hub. For example, Ethernet specifies a maximum cable length of 100 meters. At most, a wire should only run 90 meters from the hub to the wall drop.
The network installation process.
While measuring the distances, take a chance to visually inspect all the locations where running wiring for any potential problems such as electrical power sources or unusually thick walls. Look for places to run the wiring, ideally through a dropped ceiling, cable trays, and previously drilled holes. If the building does not have dropped ceilings available, and has not already installed cable trays, then it is recommended that the district install cable trays for the wiring. These are simply plastic or wooden trays attached to the walls or ceilings in which to place wiring. Their advantage over running the wiring in conduit is that they allow easy access to the wiring if a problem occurs, and can easily accommodate additional wiring in the future. Look for locations where other accessories such as base plates, face plates, wire mold (for containing vertical wiring), and raceways (for containing wiring running across a floor) will need installation. The designer should note these also and prepare them for installation as well.
A district should be sure that it follows all the local and state fire codes for a school building. In many cases, this will require the use of plenum rated wiring, which although more expensive, does not produce any toxic fumes when it burns below a certain temperature. A qualified electrician or architect can provide fire code information.
Another problem that may appear is the use of multiple power feeds into a building. This can happen if a building expands multiple times over many years. If the feeds to the building come from different transformers, the ground voltage on each feed may not be exactly the same. This can lead to problems with electrical equipment connected across the differently powered parts of the building, potentially destroying computer equipment. If the situation exists, then the use of specially insulated equipment or the use of fiber cable (which does not conduct electricity) is recommended. If the district is unsure about the power of the building, it should contact a qualified electrician or representative of the power company for more information.
3.1.2 Implementation tips
If a district is going to attempt to install the network using volunteer or previously untrained staff, the person who will be leading the process should receive extensive training that they can share with the others. This may include talking to qualified electricians and getting training from other schools that have already done the installation. For a district considering installing the network themselves, Table 3-1 shows a partial list of the equipment needed. As talked about in the section on network implementation choices, some of the wiring requirements are very strict about where the wiring can and can not be run. A district should be sure to understand and follow all the requirements. A failure during this part can render the entire network unusable, or even worse, cause intermittent, untraceable problems.
While trying to pull wiring throughout a school building, a frame or other device from which to pull the wire will need to be built. Most wire is delivered on 1,000-foot rolls. An ordinary 2-wheel hand truck can function as a relatively compact unit from which four standard spools can be pulled simultaneously. This unit can then double as a means for moving wire about the building.
In cases (very common in most modern schools) where the network cable is pulled through drop ceilings, some type of twine, lead, line, or fish-tape needs to be used. A wide variety of such materials is available. Mahomet used 70-pound test nylon masonry twine with success. With a weight on the end, the twine is light enough the designer can toss the twine 20-30 feet horizontally through a drop ceiling and run no risk of snapping it when the cable is pulled.
The designer should label all wires in the network according to a logical and clear code. If possible, they should incorporate any existing building space designations into the code. They should place this code on the physical network in three places: on both ends of each wire and somewhere in the base plate box to which the wire is connected. Copies of this wiring code should be deposited with the school district central office, the building office affected, and with those in charge of maintenance.
There are many issues to keep in mind while wiring the network. The district should try to follow all of them, but if problems or questions arise, remember the most important one: do not hesitate to get help from someone more qualified, even if it costs money. It is better to spend the money now, rather than on having someone come in later and fix a network that does not work.
3.2 Local area networking
Each building in a district will have a separate LAN (see Chapter 2 for more information about LANs and WANs), allowing each room in a building to be connected with all the others using a high speed network. There are many ways to connect the building, using different types and brands of equipment. Although each brand offers slightly different features, most will sell the same basic pieces of equipment. This section discusses most of common types of equipment. If it does not discuss a piece of equipment, or the district is unsure which type to use, do not hesitate to speak with a vendor about what products fit the districts needs. Although the vendor will often try to sell their brand, they can usually offer good advice as well.
3.2.1 Network implementation choices
The district will need to make a decision about what physical protocol to use. Currently there are several in use, although Ethernet is the most popular. Others include Token Ring, FDDI, ATM, and LocalTalk. Each protocol is capable of a different maximum data speed, known as the bandwidth, and costs different amounts. Table 3-2 summarizes all of this information. Looking back to the OSI model in Figure 2-4, all of these protocols except ATM exist at the bottom two layers, the physical and data link layers. ATM exists at those layers as well as at the network level.
Ethernet. Ethernet is an industry standard protocol operating at 10 Mbps that is currently in wide use. The protocol uses a principle called Carrier Sense Multiple Access/Collision Detection (CSMA/CD) which has two important parts to it. The first is that it is a multiple access protocol allowing all the machines to share the same physical wiring instead of requiring separate wiring for each machine (except in the case of a star topology). The second is that it operates on collision detection. Since many machines share the wire, two machines may try to use it at the same time. If this occurs, a collision has occurred. The network hardware detects the collision and aborts the network access. After a small random delay, the hardware tries to transmit again. The result of this protocol is that on very busy networks with many machines, a large number of collisions can occur, wasting a significant amount of time retransmitting information. This is why there is a recommended limit of 25 to 30 machines on a single Ethernet network.
Wiring that uses the Ethernet protocol comes in four different physical varieties: thick, thin, twisted pair, and fiber-optic. Thick Ethernet, also known as 10base5, is the original wiring used for Ethernet. Networks using it connect in a bus topology. It is the second most expensive of the four types, but also has the second longest maximum distance of 500 meters. It is not often used today because of the cheaper alternatives. Thin Ethernet, also known as 10base2, coax or cheapernet, is also a bus topology. It runs on 50 ohm coaxial cable and often connects small networks. It suffers from the problems discussed under bus topologies and the Ethernet specifications limit its maximum length to 200 meters. Twisted pair Ethernet, also known as 10baseT, runs over category 3 (cat 3) phone wiring or better. It connects in a star topology, although it shares the same CSMA/CD protocol as the other Ethernet varieties. It has a maximum length restriction of 100 meters. Fiber Ethernet, also known as 10baseFL, is not normally used except to connect hubs over long distances. Its maximum distance of 1,000 meters makes it ideally suited for this type of job.
To avoid problems with the actual electrical signal that propagates across the Ethernet wiring, the "3-4-5 rule" exists. It states that between any two machines on the Ethernet network, there must be at most five wiring segments, four repeaters connecting the segments, and only three of those segments can have workstations connected to them. A district most often will violate this rule when a repeater connects a workstation that is beyond the distance limit for the network, or when classrooms have their own hubs.
Fast Ethernet. This is a new enhancement of Ethernet that runs at speeds of 100 Mbps, ten times the rate of original Ethernet. Known also as 100baseT, it requires that the wiring it runs over be category 5 (cat 5) wiring, a higher quality than the cat 3 used by normal Ethernet. The equipment needed to use Fast Ethernet is also more expensive than normal Ethernet, and although the prices are dropping, it is most likely too expensive for schools to install initially. Like 10baseT, it connects as a star topology and has a 100 meter maximum length restriction. This allows a district using cat 5 wiring to begin with 10baseT and later upgrade to 100baseT without replacing the wiring.
Token Ring. IBM developed Token Ring. It is a Carrier Sense Multiple Access/Collision Avoidance (CSMA/CA) protocol. It passes a theoretical "token" around the network. Only while a machine has the "token" can it send information. Since there is only one "token", this preventing two machines from broadcasting at the same time. It operates at both 4 and 16 Mbps. It is not widely used because the performance increase does not outweigh the difficulties and prices required when installing a ring topology.
FDDI/CDDI. Fiber Distributed Data Interconnect (FDDI) is a 100 Mbps fiber-optic based network. Like most other technologies based on fiber, it requires two fiber cables, one for transmitting and one for receiving. The cost of fiber cables makes this choice significantly more expensive than using Ethernet. Its advantage is the higher speeds it offers, although with the availability of Fast Ethernet, this is not as significant of a factor. Ordinarily, FDDI creates a fiber "backbone" that connects to all of hubs in a large building or campus. Copper Distributed Data Interconnect (CDDI) is a proprietary variation of FDDI that runs over cat 5 twisted pair.
ATM. Asynchronous Mode Transfer (ATM) is a new technology that is in the transition from lab research to commercial use. It offers speeds beginning at 45 Mbps and can increase to even higher speeds. It runs over cat 5 twisted pair and fiber-optic cables. In the future, as it becomes more available and prices drop, it will become a viable upgrade option.
LocalTalk. LocalTalk is the original network hardware that Apple Computer shipped with its Macintosh and Apple II series computers. It is not the same as Appletalk, the network protocol used over any physical network. LocalTalk has a maximum speed of 230 Kbps, significantly slower than any of the other protocols. It is a bus topology with a maximum distance of 300 meters. Using Phonenet equipment, it can run over standard cat 3 phone wiring. With its speed limitations, it is not recommended for use in any schools unless already installed and then a district should consider an upgrade.
Figure 3-2 shows a comparison of the bandwidths offered by the different protocols, as well as different WAN technologies discussed in the next section. The figure is logarithmic and each horizontal label is ten times faster than the one to it's left.
To allow for future growth to higher speeds, the Electronics Industry Association/Telephone Industry Association (EIA/TIA) has recommended the use of only cat 5 wiring in all installations because of its capability to run at higher speeds than cat 3. As previously mentioned, this will support future upgrades to Fast Ethernet or ATM.
The bandwidth of different network technologies.
3.2.2 Equipment needed on a workstation
Once a district decides on the network's physical implementation, they will need to consider what equipment is necessary to connect all the workstations in the building to the network. Workstation is another term for a networked computer, whether the computer is a Macintosh, a Windows machine, or any other type of computer. This will normally require a network interface card (NIC) for each machine, although some new machines are now shipping will the network card built in to the computer. Depending on the network choices made, the district will need to buy the card with the correct connector, either a coax, RJ-45, Attachment Unit Interface (AUI), or fiber.
If a district is using Ethernet, the most likely implementation, they will have many choices. Since Ethernet comes in several varieties, some Ethernet cards, often called combo cards, come with two or three connectors as shown in Figure 3-3. If a school is currently using one variety and is planning to upgrade to another, then combo cards can save money. If the network is going to be exclusively twisted pair, then there is no reason to spend the extra money on a combo card with extra, unused capabilities. Another option that is appearing is combo 10/100 Ethernet cards that can run on 10baseT and 100baseT networks. The district should consider how quickly it will upgrade other parts of the network to support Fast Ethernet before investing in these cards.
The connectors on a Ethernet combo card.
The AUI connector on an Ethernet card allow easier expansion on the card. Devices connect to it, called transceivers or Media Access Units (MAUs), which offer a connection to all four of the different types of networks. Although this is redundant on combo cards, it can be useful if the card does not have an RJ-45 connector and the school is upgrading to 10baseT. Similar in function, although incompatible in size and shape, Apple Attachment Unit Interface (AAUI) connectors exist on the motherboards of most Macintosh models currently available. Be sure to buy the correct AUI or AAUI connectors when buying the network cards. Finally, do not forget to buy the necessary cabling to connect the network card if it does not include the necessary cabling.
3.2.3 Network equipment
Equipment used to create the network infrastructure falls into two categories, passive devices and active devices. Passive devices do not usually affect the capabilities of the network and make the installation or maintenance of the network easier. These devices include patch panels, patch cables, fiber boxes, and fiber jumper cables. Active devices connect the different workstations and other active devices on the network. They include repeaters, bridges, switches, and routers and the following sections discusses them in detail. A typical network will look similar to Figure 3-4. Table 3-3 at the end of the section lists a summary of all the network equipment.
A typical school network.
Patch panels. All the wiring that comes from the classrooms needs to connect to the hub. A patch panel is often used as an intermediary. All the wiring is "punched down", or attached, to the back of the patch. The front is made up of RJ-45 connectors into which patch cables connect. This prevents any damage to the room wires from occurring since changes do not have to modify the room wires.
Patch cables. Patch cables, or jumper cables, connect the patch panel to the hubs, connect workstations to the wall jacks, or connect multiple hubs. This allows the network administrator to easily reconfigure the network since all that needs to modified is which jacks the patch cable connects. The first two types of connections require a straight through cable. The third requires a crossover cable.
Figure 3-5 shows the difference between the two types of cables. The reason the different cables are needed is because of the way twisted pair networks connect. One pair of wiring transmits information from one machine to another, and a second pair transmits information in the reverse direction. A hub reverses the wiring from a workstation allowing the lines to match up correctly and a straight through cable works. However, when connecting two hubs, the wiring in both is reversed and so both would attempt to transmit on the same pair and to receive on the same pair, causing them to be unable to communicate. A crossover cable reverses the transmit and receive lines so that the hubs can correctly communicate.
A straight through and a crossover patch cable.
Fiber boxes and fiber jumper cables. Just as patch panels and patch cables attempt to prevent damage to the wiring going out to the classrooms, a fiber box and cables do the same for fiber networks. The fiber from the other part of the network connects inside the fiber box, and a fiber jumper cable is used connects the box to the hub.
Repeaters. Repeaters are the most basic type of active network equipment. They operate solely at the physical layer, receiving a signal on one port, or connection, and rebroadcasting it on all of its other ports. They can extend a network beyond the limits imposed by the wiring by boosting the signal level.
Most advertisements for "hubs" are referring to multiport repeaters. They usually come with a number of ports that are multiples of 12 and allow the network to support up to that number of workstations. Some hubs are stackable, which means they have a special connector that allows a district to easily connect more than one hub together. Others come as chassis systems for which support additional cards, each card having another 12 ports.
Switches. Switches are an advanced form of repeaters. They also act at the physical level by repeating the signal. Unlike repeaters that repeat an incoming packet out all of its ports, a switch looks at the destination of the packet and only sends it to the port of the destination. This can reduce excess traffic on a network since it isolates each port and send fewer packets to each port, thereby reducing collisions and increasing the performance of the network.
Bridges. Bridges operate at a higher level than repeaters, working at the data link layer and looking at the actual packets that are on the network. When they receive a packet, they store the entire packet in memory, verify its correctness, and retransmit it on the correct port. This allows them to connect different types of Ethernet networks together such as a 10baseT and a coax network. They also reset the "3-4-5 rule" for each port, making each port its own network. This is because it stores the entire packet and rebroadcasts it, isolating each port from the others. Like switches, they look at the destination of the packet and only send it to the port where the destination is located, reducing traffic on the network.
Routers. Routers operate at the network level. They receive a packet, view its destination, and determine if the packet is destined for a network that is directly connected to the router or if it is destined for a network further away. If it is the first it sends the packet to the correct port. If it is the latter, it sends the packet to the next router along the path to the packet's final destination. For this reason, routers typically connect between a LAN and a WAN to limit the traffic on the WAN to only packets that need to cross it. Additionally, because routers look at the network information from a packet, they can convert between different network protocols.
Along with the basic networking equipment, a district that is still using LocalTalk networks will need to install some specialized equipment that connects those networks to the building Ethernet network. This equipment includes MacLAN patch panels and GatorStars or similar equipment, or EtherPrint boxes.
GatorStar/GatorBox. If a building has many LocalTalk devices to connect, a good solution is to use a GatorStar or a GatorBox. A GatorStar connects to an Ethernet port on a hub and to a MacLan patch panel and converts the packets on the LocalTalk networks into Ethernet packets. This allows workstations on a LocalTalk network to act as if they were directly on the Ethernet network. The GatorBox is a smaller version that can connect to a single LocalTalk network and bridge that network to an Ethernet network. Please note that these are the products of a specific company and that other companies manufacture similar equipment.
MacLAN patch panel. A MacLAN patch panel is a special type of patch panel. When connecting rooms that are using LocalTalk, a patch cable runs from the normal patch panel to the MacLAN patch panel. This connects all of the LocalTalk networks to the MacLAN patch panel. A GatorStar, but not a GatorBox, requires the MacLAN patch panel.
EtherPrint boxes. If a building has only a few localtalk devices such as printers, a cheap and convenient solution is to install EtherPrint boxes next to the printers. They are small devices that convert Ethernet packets into LocalTalk packets. Unlike GatorStar or GatorBoxes, these have a limit on the number of LocalTalk devices they can connect. Please note that EtherPrint boxes are the product of a specific company and that other companies manufacture similar equipment.
3.2.4 Placing network equipment
Most vendors design network equipment to be rack-mounted. This means the equipment has special connectors on its side that allow it to bolt to a specially designed rack. This keeps the equipment off the floor, allows good air flow around all the equipment, and provides easy access to both the front and back of the equipment.
Standard racks are 22" x 36" x 84" (width x depth x height) and bolt to the floor in a permanent location. To allow easy access to both the front and back, enough clearance for a person to stand on either side should be allowed around the equipment.
Another type of rack that is useful in an unlocked room is the cabinet rack. These come in two heights, 40" and 78". They are lockable, double hinged cabinets which allow easy access to either the front or the back of equipment.
3.2.5 Configuring network equipment
Network equipment comes in two types when talking about configuring the equipment, manageable and nonmanageable. The first allows a central location elsewhere in the building, or elsewhere in the district to configure and monitor the equipment. The second requires that an administrator configure the equipment by connecting a portable computer to it. A district will need to decide if it can afford the additional costs for manageable equipment and if they will ever need the capabilities. If the district will rarely modify the network, then the nonmanageable may be a good choice. However, if a small number of people are supporting the network, their ability to remotely configure and monitor a device for correct functionality can more than make up for the cost.
3.3 Wide area networking
To provide access between the different schools and to the Internet, a district will need to install a district WAN. This will connect all the schools to each other as well as to an ISP. In some cases, the options available at the ISP will dictate the WAN technology. In others, the cost will decide the WAN technology.
When discussing costs for WAN technology, there are two separate costs associated with the network. The first is the start-up cost which includes such things as equipment and installation. The second is the recurring cost which occurs either on a monthly or a yearly basis. The start-up cost is usually higher than the recurring cost, but the recurring cost can end up being the one more expensive because a district pays it indefinitely.
The data travels over a WAN in two directions, from the ISP to a building and from a building to the ISP. The first is called the downlink because data is being sent down, or downloaded, from the ISP; and the second is called the uplink, because information is being uploaded to the ISP. On many of the WAN technologies, the uplink and downlink speeds will be the same. Several of the technologies offer different rates in the two directions. The downlink rate will be the most important because the typical use of an Internet connection is to request information from a server, resulting in requests and acknowledgments being sent across the uplink and the responses from the requests travelling across the downlink.
The different technologies include Plain Old Telephone Service (POTS), leased lines, Integrated Services Digital Network (ISDN), wireless, cable tv, satellite, and fiber. Table 3-4 summarizes the technologies.
3.3.1 Plain Old Telephone Service (POTS)
Just as its name implies, this technology relies on standard phone lines to connect schools. Using current analog modem technology, this service can run at speeds up to 28.8 Kbps in each direction. It is a cheap and usually easy way to get a machine connected to the Internet quickly. It has low start-up costs and low monthly rates. If an extra phone line is already available, the school can reuse it with no installation charge. An office can even share the line because it is only connected when it is in use, although this is not recommended. An ISP will need to provide Internet connectivity, but these are usually available at reasonable rates as well. As shown in Figure 3-6, there is no additional equipment needed except for a modem on each machine that requires Internet connectivity.
A POTS dialup WAN connection.
3.3.2 Leased line
Another type of WAN connection is a leased line. With a leased line, the district buys a permanent connection from the phone company, either to another school or to an ISP. The district sets up a router and a Channel Service Unit/Digital Service Unit (CSU/DSU) at each school that will connect and connects the leased line to the CSU/DSU as shown in Figure 3-7. This establishes a network between the two ends of the leased line.
There are two varieties of leased lines, and both operate at identical uplink and downlink speeds. The first is called a dry line, has a maximum speed of 56 Kbps, and a maximum range of 2-4 miles along the length of the phone wiring. The phone company does not provide any boosting of the signal and this limits the length of the wiring. However, because the phone company does not have to provide any equipment, the lines are available at about standard phone rates. The installation is somewhat expensive, being about $500-$1,000 for the initial setup. These lines are perfect for connecting between closely situated buildings.
The other variety is simply called leased lines. They are available in speeds ranging from 56 Kbps up to and beyond 1.5 Mbps (a T1 line). They require the same equipment as a dry line, but the phone company boosts the signal along its path. This allows leased lines to run almost any distance, although longer distances will cost more. The installation costs are similar to dry lines, but the monthly costs are substantially higher, beginning at around $200/month for short distances for a 56 Kbps line.
Connecting the district in a star topology, with only one building acting as the central hub and connecting to the ISP, is the cheapest way to create the WAN. However, because all the schools will share the same connection to the Internet, performance will eventually suffer if the shared line is not fast enough. A district will obtain the best performance by connecting each school directly to the ISP, but this will cost substantially more.
Along with connecting the district buildings, a connection needs to run to an ISP. This can cost a large amount of money, depending on the ISP. Costs depend on the speed of the connection, but commercial rates begin at about $250/month for a 56 Kbps line and extend into the thousands of dollars per month for T1 speeds. Obviously, a non-commercial rate to be found to be affordable for a school district. See section 2.3 on for suggestions about locations to contact for better rates.
A leased line WAN connection.
3.3.3 Integrated Services Digital Network (ISDN)
ISDN is a technology that is finally becoming available from most telephone companies. It offers connections at either 64 Kbps or 128 Kbps in each direction. It also expands by multiples of 64 Kbps. The basic service, called a Basic Rate Interface (BRI), contains two 64 Kbps data lines called B-Channels and a 16 Kbps control line called a D-Channel.
Unlike leased lines that are paid for 24 hours a day, 7 days a week regardless of their use, the phone company meters ISDN lines like a standard phone line. They only bill a district for the actual usage time of the line. Most ISDN connections will close after a few minutes of inactivity, keeping the usage charge down to a minimum. In addition, unlike a normal POTS modem, an ISDN connection is usually "dial on demand." This means that if a need arises to use the connection, the ISDN equipment will automatically establish the connection within one or two seconds. This allows users to continue working normally as the line connection opens and closes automatically. As shown in Figure 3-8, the actual implementation of an ISDN WAN is very similar to the network used in a leased line network. However, an ISDN Network Terminator, Type 1 (NT-1) connects to the phone wiring instead of a CSU/DSU. Many NT-1 boxes even contain a simple router so the district can save money by not needing a router at each building. For more information about ISDN, Dan Kegel's ISDN web page at http://alumni.caltech.edu/~dank/isdn/ is an excellent reference.
An ISDN WAN connection.
3.3.4 Wireless
With the phone company charging on a monthly basis for any network services it provides, other options apart from the phone company may be cheaper. One approach is to use wireless communications to connect multiple buildings. Offering high speeds and minimal or no recurring costs, these technologies deserve a good look. When compared with the phone company solutions, the up front costs are high, but over time this cost can amortize to the point where it is cost effective. Wireless may require expensive towers to clear natural and man-made obstructions around a school because all wireless communications require a clear line of sight between the transmitting and receiving equipment. Figure 3-9 shows the basic configuration of a wireless network.
Wireless technology uses one of two methods of communication, either lasers or microwaves. Laser technology offers higher speeds, but shorter ranges of only a few hundred meters. A more significant problem is that most lasers operate in the infrared spectrum and most conditions that block visible light such as rain, fog, or any other physical obstruction also block the laser. Due to these problems, lasers are not particularly useful when trying to reliably connect multiple buildings.
Microwave communications are, on the other hand, immune to many of the problems that plague lasers. They are immune to most weather conditions and can travel much farther distances.
Most microwave equipment that a district will use operates in one of two freely available unlicensed spectrums of the radio frequency band. The advantage is that it requires no licensing with the FCC. Unfortunately, this also means that other equipment can also use the same frequency and can cause interference. However, the equipment uses techniques to avoid this interference from causing problems with the information being send across the connection.
Although many products exist in the wireless market, two specific products can be used as an example of the capabilities of wireless networking. The first is the Wireless KarlBridge from KarlNet and the second is the AirLAN Bridge/Ultra from Solectek.
Wireless KarlBridge. The KarlBridge offers a maximum data rate of 2 Mbps in each direction and has a maximum range of 10 miles. It costs about $6,000 for a complete setup, including two antennas, two bridges, and all the cabling needed to connect the equipment together. Although not currently in use by any school district, the University of Illinois installed and tested the equipment and found it to meet its stated capabilities. More information about the Wireless KarlBridge is available over the phone at (614) 263-5275 or on the KarlNet web site at http://www.karlnet.com/.
AirLAN Bridge/Ultra. Offering longer ranges of 10 to 25 miles, the AirLAN Bridge is a choice for district who need to connect to points farther away. It also operates at 2 Mbps in each direction. The entire setup ranges in price from $10,000 to $15,000 depending on the distance needed. It includes two antennas, two bridges, and all the cabling needed to connect the equipment. Further information about the AirLAN Bridge/Ultra is available at (800) 437-1518 or from the Solectek web site at http://www.solectek.com/.
So far, Mahomet and Tolono have mixed results using wireless technology. Tolono was able to get their connection up and running on the first try, and it has run almost perfectly since then. Mahomet has had considerable problems getting their connection to stay up. They have replaced the equipment several times without successfully solving the problem. The vendor, Solectek, has been extremely helpful during this period and has replaced the equipment at no charge and has even paid for the tower climbs needed to replace the equipment. They are currently working on a solution to the problems that the district is having and feel confidant that they will be able to solve all of Mahomet's wireless problems.
A wireless WAN connection.
3.3.5 Cable TV
Cable tv is another alternative to the phone company that is emerging. It allows a district to provide WAN connectivity using the cable tv wiring. Although still a new technology, it promises to offer high speeds and low costs. However, several technical problems need solutions. These are due to the current implementation of cable tv systems that send the same signal from a central office to many destinations.
The first problem is the fact that television broadcasting is inherently a one directional process. Information is not sent back to the central office when broadcasting television. New cable equipment that allows two-way communication is now available and as the cable company installs upgrades to the cable tv equipment, this problem will be eliminated.
The other problems that exist are based on the fact that many buildings share a cable tv signal. This leads to the two problems of shared capacity and security. Although cable tv networks operate at high speeds, many buildings share the network. This results in each building getting only a small part of the total bandwidth. Security is also a problem, when the sharing information among several buildings, because someone with the correct knowledge and experience can view any unencrypted information placed on the network.
Currently, two implementations of cable tv equipment are available, and Figure 3-10 shows an example of each. The one available from local cable companies depends on the capabilities of the cable tv system currently installed. In the first implementation, a cable tv line runs to the building and connected to a cable modem, and the cable company sets aside a single television channel as the downlink. For the uplink, a telephone modem transmits the information. This implementation is known as an asymmetric or hybrid solution because of the different paths taken by data traveling in different directions. This solution does suffer some serious performance problems. Using a television channel, the downlink has a maximum rate of 4 Mbps, but the POTS modem limits the data being sent across the uplink to a rate of 28.8 Kbps.
The second implementation requires upgraded cable equipment that can support bidirectional signals on the cable network. These networks, known as symmetric solutions, use two channels for data, one for the uplink and one for the downlink. This allows data to travel at 4 Mbps in both directions.
In terms of costs, the cable company will most likely want to charge a monthly fee for each building that connects. Some cities, such as Glenview, IL, have successfully avoided most or all the cost by requiring data connections to all the city buildings, including schools, in the contract negotiated with the cable company.
A Cable TV WAN connection.
3.3.6 Satellite
For districts located in remote areas, where other technologies are prohibitively expensive or unavailable, a satellite connection is a viable alternative. With this technology, a district buys a satellite dish and the necessary hardware needed to connect it to a network. A building then has a 56 Kbps downlink connection. As in the hybrid cable tv solution, satellite is an asymmetric solution and the uplink rate is only 9.6 Kbps. The initial equipment cost is about $15,000 for the satellite dish and equipment, and the monthly rate is about $800/month. A faster option is also available, with the downlink running at 115 Kbps and the uplink running at 19.2 Kbps, for about $1200/month. These rates are high when compared with other solutions, but in some rural areas this may be the only option.
The equipment is available from Intellicom. Besides providing the equipment and network, they act as the ISP. This can save a district money that they would normally need to connect to the Internet. Additionally, aware that the uplink rate is insufficient to support a district web server, they provide space on their web server for the district to use. They can be contacted at (510) 353-0102. They also have information about their products available on their web site at http://www.vsat.com/.
A satellite WAN connection.
3.3.7 Fiber
The ideal solution for a district is to connect all of its buildings with fiber-optic cables. Fiber can run at speeds beyond 1 Gbps, allowing a district to connect the WAN at a high speed immediately and to upgrade to even faster speeds in the future. Unfortunately, this can be very expensive. Fiber suitable for use outside costs between $2/foot and $2.50/foot. When installing thousands of feet, this can become a significant cost. Besides the fiber cost, installation can also be very expensive. A district needs access to all the land between the buildings, and this "right of way" can often cost a lot of money.
When the district has multiple buildings on the same property, obtaining access is not usually a problem. A district either needs to bury the fiber in the ground, requiring expensive excavation, or hang it from telephone poles, leaving the fiber open to the weather. If the district chooses to lease a fiber from the local telephone company, then the district will have to pay a high monthly rate for the fiber.
3.4 Dialin service
A district will need to make a decision regarding the installation of a dialin pool of modems connected to the district WAN. These modems could be for staff and student dialup access to the district WAN and/or to the Internet. This may seem like a good service to provide to the district. However, with most phone local and long distance phone companies offering this service for only about $20/month, it is not a service most districts should consider installing. It can easily become an expense and management problem. As more people begin to use the dialin service, a large percentage of the modems will show a significant increase in their use. A district would then need to buy additional modems to provide more dialin connections. A district could spend the money used for the modems and phone lines elsewhere with better gains to a district.
3.5 Software products
Along with installing a network, a district also needs to purchase or download from the Internet any software products it will need. Without the correct software, a district will not be able to access or offer the services that it decides are necessary. These include driver software such as TCP/IP drivers, application software such as email programs or word processors, and server software such as file servers or a web servers. Figure 3-12 shows how the different network software pieces interact on a Macintosh and a Windows machine. At the lowest level of a machine, a hardware driver communicates directly with the network interface card installed on a machine. Software drivers supporting the different software protocols such as AppleTalk, IPX, and TCP/IP act as intermediaries, communicating with both the hardware drivers below them and the applications and servers above them. Applications and servers provide an interface to the user. Just like in the OSI model, a layered approach is used. This insulates the upper layers from the requirements of communicating with the specific hardware present on a machine. This allows all application and server software to function identically regardless of the brand of equipment installed on a machine.
The software installed on a workstation.
3.5.1 Driver software
There are two parts to the driver software as shown in Figure 3-12, a hardware driver and a network driver. The hardware driver usually comes on a disk with the network card, along with instructions on how to install it. Some of the network drivers come pre-installed on computers, while others need to be bought. On a Macintosh, the AppleTalk drivers have always been a part of the standard system software. The MacTCP software, the TCP/IP driver offered by Apple, was not part of the standard system software until the release of System 7.5 several years ago. Before that time, MacTCP was a separately purchased product. All machines shipped over the last few years, and all currently shipping machines include MacTCP, but any machine that is not running System 7.5 will require that a copy of MacTCP be bought for it. Of all the Windows varieties, only Windows 3.1 and earlier do not include network drivers as part of the standard system. It requires an additional product be bought which will add a WinSock interface to Windows, allowing all TCP/IP based applications and servers to run. Windows for Workgroups, Windows 95, and Windows NT all include the necessary network drivers as part of the standard system software.
3.5.2 Application software
Applications are programs such as Microsoft Word or ClarisWorks that run on a workstation. They allow a computer user to perform a task on the computer such as write a paper, check their email, or browse the Web. When selecting applications, a district can establish a standard that all the buildings in the district must follow, or they can allow each building to choose their own. By establishing a standard, a district can assure staff and students access to their files be from anywhere in the district. A standard will also make training and support easier since the entire district will only use a controlled group of applications. Without a district standard, each school can choose applications that their staff is already familiar with or which serve a specific purpose such as multimedia authoring. A district can also select different standards for different schools, depending on the grades taught at the building. The best approach for a district is to establish a standard for everyday applications such as word processors or email programs, but to allow freedom when choosing special purpose software. This allows building staff to choose software to fit their special needs while still making papers and other common files easily shared among buildings.
Applications fall into two categories, stand-alone and clients. The first are programs such as those listed in Table 3-5, which do not require interaction with other machines to function. They are available from many vendors, offering different capabilities and costs. A district should compare all the alternatives for a given type of program, and select the one offering the best price and feature match. Districts should consider buying integrated packages that offer a word processor, a spreadsheet, and a database in a single product. Districts should also buy only the capabilities they need or expect to need in the future. For example, if a district is going to select an integrated package as a standard, it should consider if it needs the extra capabilities of Microsoft Office. If not, the district can save money by selecting ClarisWorks or Microsoft Works instead.
The second group of applications are ones such as those listed in Table 3-6, which require a network to function. Many of the applications in this category are available as either shareware or freeware, and offer similar or better features than the competing commercial products. A district should investigate the free alternatives before it spends money on commercial products. In addition, the choice of client software may depend on the server software selected. For example, if district decides to standardize on Internet email, then they must choose an email client such as Eudora that supports Internet email.
3.5.3 Server software
Server software runs on a machine and provides the services Chapter 2 discussed. Often server software requires a dedicated machine for it to run on, and even when it does not require a dedicated machine, server performance is usually better if applications are not run on the machine. The server software will also run faster if installed on a faster computer. For this reason, a district should plan to buy separate, fast machines for use as servers. Server software exists for both Macintosh and PC platforms, as well as for Unix machines. Unix machines are complex, high performance workstation often used in the engineering world. Most of the platforms support all the services as shown in Table 3-7, although some brands of server software will only run on specific platforms.
File and print server. There are three main choices when selecting a file and print server, AppleShare, Windows NT, and Novell Netware. As shown in Table 3-8, they offer many of the same features. All of them support both Macintosh and Windows 3.1 clients, but if a district is using other clients, such as Windows 95 or OS/2, they will need to select an appropriate server that can support those clients. The installation and maintenance of the file server software are also issues that needs consideration. File servers are complex packages requiring training to be used optimally. If a staff member already has experience with one of the choices, then a district can save time and money by not needing to train a server administrator on a new product.
Multimedia server. Multimedia comes in a variety of forms. It includes audio and video segments often on multimedia CD-ROMs. Depending on the needs of a district, the same platforms acting as the file server can also act as the multimedia server. All three platforms support the sharing of multiple CD-ROMs. Additional software can be bought for Windows NT and Netware that will allow them to serve audio and video segments. For serving large numbers of audio and video segments, the optimal solution is to use a Unix system with software specifically designed specifically to serve these segments. This can cost a lot of money, and requires the district to support a Unix system.
Email server. Selecting an email server can be a complex problem. Each brand, such as Microsoft Mail or QuickMail, uses its own mail protocol. The Internet also supports several different protocols. All of these email protocols are incompatible with each other without software to translate between the protocols. Server programs, called software gateways, will translate between most of the different email protocols, but a district can save money and avoid problems by standardizing on a single email protocol. The recommended email solution combines two protocols used on the Internet, Simple Mail Transfer Protocol (SMTP) and Post Office Protocol (POP). The first transmits email from the sender to the receiver's email server, and the second allows the receiver to view their email. As shown in Table 3-9, server software to support the Internet protocols is available for all three file server platforms, and the file server can also act as an email server. As an added benefit, all the software listed in Table 3-9 is available at no cost. Unfortunately, these email servers can not handle a large number of users, and the performance of the servers decreases after several hundred email accounts are in use. A Unix machine act as an email server for a district requiring better performance, although this is not recommended unless a district already has a Unix machine running for another purpose or absolutely requires higher performance.
Web server. Web servers are also available for the same three platforms that file servers and email servers can run on, and can often share the same machine used as a file and email server. Table 3-10 lists some of the choices available. The wide range of choices allows a district a lot of freedom in selecting a web server. Some issues to consider while selecting a product are cost, performance, and how easily and in what programming languages a user can expand the web server. Some commercial products can cost up to $1,000. Performance is also an issue, but this is tied to the price. The more expensive the web server, the better performance it usually offers. An exception to this is the Netscape web server, a high performance product, which is available to educational users at no cost. The last consideration is how the web server offers access to external programs. These programs, called Common Gateway Interfaces (CGIs), handle the advanced actions on a web page such as forms processing. Each platform offers a different interface to CGIs, and CGIs designed for one platform are not usually usable on another. Again, a Unix machine can act as a web server for a district requiring better performance, although this is not recommended unless a district already has a Unix machine running for another purpose or absolutely requires higher performance.
Usenet news server. Currently, only Unix systems can run news servers. If a district needs to make Usenet available to its students and wants to maintain its own server, then it will require that a district buy and maintain a Unix system. This can be a complex process, especially when trying to run a news server on it. If a district is willing to let another group control its news server, the ISP will often offer the service, allowing a district to access the news server run by the ISP. Although not a perfect solution, this prevents a district from having to maintain its own Unix system.
Administrative record keeping system. A district usually buys a n administrative record keeping system as a complete system containing all the necessary hardware and software needed to install and use the system. It will include both server and client software in the package. Unfortunately, the complexity involved in managing student records and grades can often result in a cumbersome program that is difficult to use. A district should keep this in mind when selecting a system and be sure that it is easy to use. A district should also consider the performance of the product. Although it is impossible to give exact performance numbers, the system should not make an administrator wait to retrieve information. The system should also be customizable, allowing changes in its reporting and record keeping capabilities without requiring a new system to be bought.
Most administrative systems operate in one of two environments, either centralized or distributed. In a centralized environment, a central computer stores all the information is, often at the district administrative office. All requests for information travel across the district WAN to the central office where the system fulfills the requests and sends responses back. The system generates all grade, attendance, and other reports at the central office and then the district staff distributes printed copies as needed.
In a distributed environment, a student's building stores their records. When another building needs to access the information, it sends a request across the district WAN to the student's building and the systems at that building fulfills the request. All of this happens transparently to the administrator who does not need to know the student's building. The system generates reports at each building, allowing quicker distribution of the information.
Most of these systems will cost in the tens of thousands of dollars. With that large a price tag, a district should be sure that the package they are purchasing includes technical support. If not, the district can add that in as a negotiating point when pricing the system. Additionally, the district should verify the quality of the company's technical support group, and get references of other districts that are using the product. This will ensure a district that it will get prompt and useful help if problems should arise. This is vital because once the district puts all the records into the system, if it stops working they will all become inaccessible.
Another option that a district can consider is developing a custom administrative record keeping system. This would require a district hire a full time computer consultant to design, implement, and support the system. The software that the consultant develops should meet all of a districts needs exactly. When comparing this with the ten of thousands of dollars required to buy an administrative record keeping system, this option can be competitive costwise and a district will get exactly what they want.
3.6 Cost estimates for equipment
While developing budgets, a district will need cost estimates for the district network. Some of the costs, such as the prices for WAN equipment have already been mentioned. This section details other costs. These costs, only estimates, are based on current prices, and can fluctuate rapidly.
3.6.1 General networking costs
As a rough estimate, Champaign, Urbana, Mahomet, and Tolono have found that the equipment costs for networking a small elementary school with several hundred students are about $15,000. A middle school with 1,000 to 1,500 students will cost about $20,000. A large high school with 2,000 to 3,000 students will cost close to $30,000. Another way of estimating the total cost of a building is to use a cost of $250 per drop for all the equipment needed, and multiplying that by the number of drops installed in a building. These prices only include the initial equipment and wiring needed to connect all the rooms in the building to a central hub. They do not include labor costs or the costs associated with buying computers for use on the network.
3.6.2 Network equipment costs
Network equipment prices are dependent on the type of equipment being bought, the brand and if the equipment is manageable. Simple equipment like repeaters may only cost a few hundred dollars, while a router may cost five or six thousand dollars. Manageable devices are also more expensive, adding about 30 percent to the cost of a hub, bridge, or other piece of equipment.
3.6.3 Labor costs
The labor costs will be one of the largest costs involved in the project if the district contracts an outside service to install the wiring. Even if the district decides to install the wiring using district staff or volunteers, the wiring should only be installed once. The wiring only costs about $100 for a 1,000 foot spool, and is not a major part of the cost. Therefore, if a district foresees needing additional drops in a room in the future, it should consider installing extra wires to each room at the time of the original installation. They do not have to connect until needed and only add minimally to the total cost while providing extra growth capabilities to the network.
3.6.4 Computer costs
A good estimate for the price of a new computer is about $2,000. The capabilities of the system will increase in the future, but a system that a district would want to purchase will always be priced around that range. This is for either a Macintosh system or a name brand PC-compatible system such as one from Dell or Compaq. More powerful systems for use as servers will often fall in the $4,000 to $5,000 range and having faster processors, more memory, and larger hard drives.
3.6.5 Software costs
The prices for application software will also vary, but a simple integrated package such as ClarisWorks or Microsoft Works will cost about $100 for each machine when bought in quantity. Other software will vary widely in price, with many products being available as freeware at no cost, and other specialized products, such as multimedia authoring tools, being in the range of $1,000 to $2,000 for a single copy.
These prices include a standard educational discount. If the prices a district receives from a vendor are higher, contact them about educational pricing.
As always, prices will also vary from one vendor to another. A district should be sure to compare pricing from different vendors and find the best pricing. As discussed previously about where to obtain help, the best solution is to form a partnership with a local vendor that can result in better prices.
3.7 Problems that can occur during implementation
Even after a district has created all of its plans, the network designers need to stay involved. Problems will always occur, prices will always change, and timetables will often slip. By staying involved with the implementation process, the designers can solve these and other problems before they become disasters.
By working carefully, Champaign and Urbana have been able to finish below their expected costs. Some of the reasons for this were that the designers followed the implementation process closely and were able to avoid buying some equipment that was initially planned but was later found to be unnecessary. Both districts found that the actual installation of the wiring took much longer than expected. This was because the staff involved in the installation was unfamiliar with the networking process, and because unexpected problems arose. In one building in Champaign, for example, the floors in adjacent rooms did not line up as were indicated on the floor plans. One was higher than the other by several inches, so when they drilled a hole between the rooms from the lower room, it never came out on the other side. They solved the problem eventually, but they lost time in the process.
Other delays also can occur. If a district is using an outside contractor, it should be sure to put a due date on the completion of the installation to avoid potential slips. When using volunteers, a district should be sure to invite extra people since inevitably some will be unable to show up at the last minute. Even hiring new, full-time staff is not foolproof as Champaign discovered when one of the two people hired to install the network quit unexpectedly.
Even after the network installation, delays will occur. Equipment can arrive from the vendor nonfunctional and needing to be replaced. Hardware and software will need to configuration, a process that can take an enormous amount of time. Software may not be compatible with the system it was intended to run on, requiring a shuffling of equipment.
The connection to the Internet that the district is planning can also take longer than expected to be ready for use. As previously mentioned, Mahomet has had problems with its wireless connection that prevents it from reliably connecting to the Internet. Getting a connection established through the phone or cable company can also take longer than expected, especially if the technology that the district is using is new to the utility company.
The key to solving the problems without losing too much time is for the network designers to play an active role in the entire process. If delays occur that are beyond the control of a district, then the district can redirect its efforts to another part of the network until they find a solution to the problem. Although this can be very difficult, by constantly modifying the process to fit current conditions, a district can finish their network both on time and under budget.
Return to the Table of Contents
This file last updated on 05/09/96 at 13:48:05.
© Copyright by Bradley H. Lamont, 1996