In general, information technology (IT) managers (those in charge of the office network) couldn’t care less about cabling; cabling represents only a small percentage of the IT budget, so it is often left to the cabling contractor to decide what cabling system to install; the IT manager has other problems: keeping up with all the hardware advances, software upgrades, training users, fighting spam and hackers, and controlling his users who had some tendencies toward surfing Web sites that were not exactly in the company’s best interests.
As network speeds increased, most IT managers and cabling contractors decided to convert the backbone to fiber, as it offered more upgrade opportunity. As users demanded mobility, cabling contractors agreed to add wireless. Many, if not most, IT managers just kept upgrading the copper cabling installed to the latest version.
But life has become much more complicated. New network applications have new demands on the network infrastructure. Applications—such as voice over Internet protocol (VoIP) and VoIP over wireless, 10 Gigabit Ethernet and Internet protocol (IP) digital video, closed-circuit television (CCTV) video, and other building systems on structured cabling, along with increasing security issues—have changed the playing field. Let’s consider them individually before we look at the overall network.
Since every personal computer already has a port for Ethernet on unshielded twisted pair cabling (UTP), it is a common practice to connect the PC on the desktop with UTP cabling. Since UTP cabling is limited in distance to 90 meters of permanently installed cable and 10 meters of patchcords, that usually means that the network has to include telecom rooms to transition between the backbone and horizontal cabling to the desktop.
UTP cabling is, as usual, in a transition between grades of performance—this time from Cat 6 to “augmented” Cat 6. Most networks that use nothing faster than Gigabit Ethernet (1,000Base-T) can still use the cheaper, easier-to-install Cat 5E, which has been around more than five years, or Cat 6. Augmented Cat 6—sometimes called Cat 6a—is not yet standardized, only available in proprietary “pre-standard” versions and costs about twice as much as Cat 6. Installing it at this time is a pretty risky business. Some vendors also are selling a Cat 7 cable, which is shielded, sometimes with nonstandard connectors required, which is probably also a bad choice.
The impetus for the new, higher-performance cables is the latest upgrade of Ethernet to 10 gigabits per second (10GbE). Ethernet has increased in speed by a factor of 10 with each network standard upgrade. The original Ethernet started at 10 megabits per second on coaxial and was adapted to a version of unshielded twisted pair telephone wire called Level 3 in Anixter’s jargon and Cat 3 with the first TIA cabling standards. With its first upgrade to 100 Mb/s, a new UTP cable with higher performance—Cat 5—was introduced. When Ethernet jumped to 1 Gb/s, an upgraded Cat 5 called enhanced Cat 5 was introduced to handle it, although manufacturers of most Gigabit Ethernet equipment now say their products will work on Cat 5.
The pitch for the new Cat 6a is that you need it for 10GbE. Cutting through the hype, you will find out that:
1. 10GbE is only intended for backbones. It is way too fast for any PC now and probably in the future—even HDTV video only needs 45 megabits per second!
2. 10GbE is probably not compatible with Cat 6 because the 10GbE transmission methods use high frequencies that cause crosstalk between cables (“alien crosstalk”), not just pairs between cables.
3. Cat 6a cables must be larger in diameter to prevent alien crosstalk, so they create a need for much larger spaces to accommodate cables, at much higher cost.
4. You can figure the cost doubles for the components, installation labor and certification testing.
Bottom line, the UTP-connected desktop is adequately served with Cat 5e. Cat 6 buys some performance edge, but at additional cost and installation difficulty.
Most large networks require a backbone to connect to local switches that control the horizontal UTP cabling to the desktop. Unlike the desktop connections—which often move with the users, causing changes in the cabling and or network management—the permanent backbone connects the main computer room to each telecom room. Network upgrades usually involve adding more users or upgrading the backbone capacity by using higher speed equipment. Fiber has become the media of choice here because it allows easy upgrades in speed by upgrading electronics, plus the small size of the cable allows installing numerous spare fibers for future use.
Fiber, therefore, is a “no-brainer” decision for the backbone. However, a network planned now should take advantage of the latest fiber technology, using “laser-optimized” 50/125 multimode fiber (also called OM3) for today’s equipment up to 10 GbE. Since fiber is cheap, backbone cables should include lots of spare fibers, about double the amount that are expected to ever be used, plus a number of single-mode fibers.
In the past, we have suggested leaving the single-mode fibers unterminated until they are needed, but that has caused problems. Some users did not remember the fibers were single-mode, leading to incorrect termination procedures or trying unsuccessfully to run multimode electronics on these single-mode fibers. Now we suggest all fibers be terminated and color-coded (beige for MM, black for SM, green for SM APC). Cables need color-coding, too, with the new laser-optimized 50/125 colored aqua, so users will not confuse it with 62.5/125 cables, which should be orange.
The connector is another technology change for fiber. The small LC connector has become the connector choice, replacing the SC and ST used most often in the past. Certainly the LC is the choice if 50/125 fiber is used in the backbone cable, as the LC will prevent using older 62.5/125 fiber patchcords with SC or ST connectors by mistake.
Any smart network manager or cabling contractor installing a new network today should evaluate a centralized fiber network as an option. Centralized fiber takes advantage of the greater distance capacity of optical fiber to go straight from the computer room to the desktop area, where an inexpensive media converter connects to the PC with a Cat 5E cable. You can even get fiber-connected mini-switches with four to eight ports that connect several desktops with one fiber, further reducing costs.
Skeptics who need convincing of the viability of this option need only price a network with and without telecom closets. Generally, the cost of space, power, grounds and air conditioning in a telecom closet will exceed the additional cost of optical electronics in new construction or where building layouts require numerous closets. If the user wants to try options in costing, there is a online cost model available from the TIA Fiber Optic LAN Section at www.fols.org.
No corporate network can probably be built without wireless today. Most laptops come with wireless connections and people want to use them, in spite of their limitations (bandwidth, number of possible users accommodated, security, etc.) Wireless is not “wireless,” as every antenna location, called an access point (AP), must be connected into the network cabling to communicate with the network electronics. The wireless connection only replaces the patchcord that would otherwise connect the user into the network. Thus every AP must have a connection into the network, either over UTP or fiber. APs are available with either type of connection and fiber versions are not that much more expensive today.
Which wireless version should be chosen? Wireless standards are developed by the IEEE 802.11 committee, and standards are already up to version “n” (802.11n). Most current networks use 802.11b or g, which offer adequate bandwidth for most users and enough channels (frequencies) to accommodate multiple access points for good coverage. It is important to note that the b and g versions are interoperable. A higher bandwidth version—“n”—is in development. It trades channel selection for bandwidth by transmitting part of the signal over one of several frequencies, similar to GbE, which uses all four UTP pairs simultaneously.
Wireless offers several challenges to the installer and user. First, it is important to provide good coverage in the work area. This involves installing multiple APs with overlapping coverage. Unfortunately, the coverage any AP provides depends on the environment, as objects like walls, office partitions, desks and even people absorb or reflect signals, affecting coverage. Manufacturers’ diagrams often look like a nice circle, while actual coverage looks more like an irregular blob. The best advice I have found is to overlap AP coverage and scan for dead spots.
Security is the next—and biggest—problem. Any AP is a potential opening for those wishing to breach your network. Any company’s connection to the Internet will have a firewall to prevent unauthorized entry by those outside the network, but any wireless AP offers easy access, even to someone sitting in a car outside on the street.
Securing wireless networks requires connecting them not to just any switch in the network, but to wireless controllers that filter all traffic from the AP to authorize users. Controllers can also provide guest access, for example allowing guests to access the Internet but preventing unauthorized access to the corporate network. Wireless controllers must also be able to identify unauthorized APs plugged into the network, as employees sometimes connect their own units, which are likely to be totally insecure.
All of the above
The typical network infrastructure will likely be configured with a combination of copper, fiber and wireless, depending on the layout, traffic and future plans. There are always so many potential options that the process of making a reasonable decision depends more on an assessment of individual user needs today and future plans than pure cost issues. EC
HAYES is a VDV writer and trainer and the president of The Structured Cabling Association. Find him at www.jimhayes.com.