A successful, cost-efficient enterprise network requires critical assessment of which type of cabling media will best suit each demanding application.
by Matt Brown
As bandwidth demands on theenterprise network continue to grow, companies are placing greater emphasis on high-performance infrastructuresolutions to address performance in various levels of the network. Today, companies have a wide variety of connectivity options to optimize network performance and management. To create a successful network in a cost-efficient manner, it is important to determine which type of cabling media best suits each application. And to determine the right media choice, you must analyze the critical requirements of each level of your network.
Making the best cabling choices
The enterprise network can be broken down into several distinct physical zones or applications: The LAN horizontal and backbone, and the data center"s horizontal, backbone and storage area network.
In a data center's dense computing environment with redundant connectivity, many ports are needed within the server cabinets, giving copper an advantage.
Each of these must be evaluated on its own to determine which cabling choice best fits each application. The cabling choices most commonly used in these applications are Category 6 and 6A UTP copper cable, or OS2 (singlemode) and OM3 (laser optimized multimode) fiber cable.
Different networking applications have different requirements and indicate different connectivitymedia. When evaluating what form of network connectivity is best suited to an application, there are five main attributes to consider:
- Total solution cost;
- Power-over-Ethernet (PoE) support;
- Distance capability;
- Port density;
- Cable density.
To fully understand each application"s importance, let"s evaluate these one-by-one.
Total solution cost includes the combined costs of the cable, connectivity, network electronics and installation. At the edge of the network, small changes in individual port cost can add on thousands of dollars to the total solution cost.
Power over Ethernet is the capability of powering remote devices with their Ethernet connection, eliminating separate power cabling. This capability has become increasingly relevant as more devices have become IP-enabled (such as IP phones and IP cameras).
Distance capability is the maximum channel length allowed by the connectivity type. For agiven data rate, fiber will always enable a longer channel length than copper cable.
Port density is the number of connections the media choice can support in a given area; typically, ports per rack unit.
Cable density is the number of channels the media choice can support in a given area; typically, channels per square inch of conduit or basket tray.
The enterprise horizontal is the network path from the employees" desktop to the telecom (TR) or network room on that floor. Total solution cost is extremely important in this application since a typical enterprise floor has from 100 to 300 ports (workstations). With fiber"s OM3 and OS2 total solution costs being more than double the cost of copper cabling, Categories 6 and 6A have the upper hand here. Adding further weight to copper"s case, fiber does not support PoE, a significant disadvantage in the horizontal where small IP devices, such as phones, cameras and access points, reside.
To meet the distance needs of the enterprise horizontal, the solution must be able to reach 50 to 100 meters to service the entire floor. While TIA TSB-155 tells us that Category 6cable might be able to run 10-Gbit/sec speeds up to 37 meters if mitigation steps are taken, Category 6A has optimal performance and is guaranteed to support 10-Gbit at distances up to 100 meters.
In a storage area network, an LC-based fiber rack can provide 567 ports, while MPO connectivity can provide 1,728 ports in 12U (six times the density of copper cabling).
As a rule, fiber has better port andcable density than copper; however, since this application involves connecting distributed work stations to a central point on the floor, cable density is not a dominant consideration. While the telecommunications room can be a crowded place, the port density advantage of fiber does not make a big difference when considering the total number of ports being serviced.
Category 6A is the most cost-effective way to provide 10-Gbit/sec to theenterprise workstations and, unlike fiber, it allows individuals to run IP phones, cameras and other devicesusing Ethernet capabilities.
The link between the telecommunications room on each floor and the building"s equipment room is one of the most highly trafficked parts of a communications network. The backbone connects each floor and each building of theenterprise, speeding information to the appropriate destination.
Wheras cost is the most important concern in the enterprise horizontal, the emphasis in the enterprise backbone is on distance capability andcable density.
While the actual distance supportrequired depends on the dimensions of the building, standards give a maximum backbone distance of 300 meters. Since Category 6A reaches a maximum ofonly 100 meters, a fiber solution becomes the clear choicefor the enterprise backbone.
The pathways connecting the telecommunications rooms to the equipment room are limited in space and have stringent fire-stopping requirements. A more dense cable solution allows for a greater number of channels inside the space-limited conduit. Fiber cable is significantly more channel-dense than copper, and so OM3 and OS2 media would be wise considerations for the enterprise backbone.
But which is the more suitable fiber choice: OM3 or OS2? While both can be used in the backbone, the total solution cost for OS2 fiber is nearly double that of OM3, making OM3 the more economical of the two.
Satisfying large port demands
In the data center LAN, the horizontal segment is the link between the server and the access or edge switch. In this dense computing environment with redundant connectivity, many ports are needed within the server cabinets. This demand for large numbers of ports ultimately drives the importance of total solution cost, giving copper connectivity an advantage.
The ability to support Power over Ethernet is not a factor in the data center, since the devices being connected (servers, switches, storage) consume far too much energy to be powered by PoE.
Connecting the large number of channels present in a data center makes port and cable density a factor in this decision, which gives some advantage to fiber; however, the use of a dedicated conveyance product, such as a basket tray and ladder rack, allows the larger coppercable to be managed effectively.
In the end, solution cost wins out, making copper the most logicalsolution in the data center, with OM3fiber a viable second choice. In choosing between Category 6 and 6A, the need for highly reliable transmission indicates 6A for 10-Gbit/sec,with its guaranteed performance to100 meters, along with its cost-efficientcharacteristics.
The data center backbone shares similar characteristics to the enterprise backbone; an edge switch is aggregating traf-fic and connecting it to a core router atthe highest speed possible. Distancerequirements are greater than those inthe LAN, with distances commonlyexceeding 100 meters. But in the datacenter, the importance of port density is greater, since the backbone will be takingaggregated traffic from many moredevices back to a main distribution area. Here, fiber provides greater distances and better port and cable densities.
OS2 fiber is used in these applications, but only where very long distances are required. Considering that port costs are significantly higher for OS2 fiber, OM3 is, by far, the dominant choice in thedata center backbone.
The storage area network (SAN) isanother network operating inside the data center. This separate, back-end network connects servers to shared storage devices. Although it is similar to the LAN in physical topology, the SAN"s performance requirements are very different.
The need for high reliability, highresilience and very low channel latency pushes SAN toward different networking applications, such as Fibre Channel, and make solution cost slightly less of a differentiator. Since storage environments can be very dense, cable and portdensity are key factors.
Here, fiber gives you access to greater cable and port densities. In a 12U rack space, for example, copper connectivity can provide 288 ports. An LC-based fiber rack can provide 567 ports, while MPO connectivity can provide 1,728 ports in 12U (six times the density of coppercabling). The combination of extreme density capability and reasonablesolution cost makes OM3 fiber the clear choice for SAN applications.
Media and management payoff
No matter what the media choice or management route taken, the broadertrend will be adding intelligence and control to the physical layer (seesidebar, page 44). Real-time management infrastructure works with yourmedia choices, whether it"s copper or fiber, and manages the system with constant updates, helping you reduce network maintenance cost, efficiently utilize assets, and increase revenue.
MATT BROWN is global data center solutions manager for CommScope Enterprise Solutions (www.commscope.com)
Managing your infastructure
Once you have the appropriate media choices, managing them is the next step in completing a high-performance network. Your selections of copper and fiber won"t be effective unless a systematic and disciplinedapproach is taken to organizing the network, specifically inthe data center environment where there is so muchequipment being interconnected.
In the beginning, IT technicians just needed the network to operate and be effective. After it was able to perform, businesses needed the network to perform faster and better than the networks of competitors. As companies grew more dependent on these networks, network management became more critical.
Within professional organizations, there has always been some level of infrastructure management, and the different ways of documenting it have evolved with time. In the 1970s, paper-based methods were dominant. Having a clipboard with a map of the cabling in your IT closet was considered "organized."
In the 1980s, companies moved to spreadsheets, which acted as the electronic clipboard. The files were portable and could be easily shared, but still required the user to make manual paper-based work orders when problems arose or changes were needed. In the 1990s, cabling management software and databases became some of the many ways system administrators kept tabs on the network. It"s easier to access the information via the network, additional information can be captured, and automatic generation of paper-based work orders can be provided.
As much as infrastructure management has evolved, however, it has still failed to provide updates that aren"t manually input. Instead of relying on paper-based or manual systems to implement and track network connectivity, real-time infrastructure management uses an integrated hardware and software system to provide true visibility and control over the physical layer of the network. With these solutions, the network manager is alerted to changes in the physical network, creates electronic work orders to provision services, and can track and report on use of the physical network assets. It lets you remotely manage the network infrastructure in real time, improving accuracy and efficiency of network maintenance.
Real-time reporting features enable the enterprise to better utilize assets. Managers can set alerts regarding the percentage of network ports being used, enabling the IT specialist to prepare for an equipment upgrade to meet the needs of the network. Constant real-time monitoring improves network security and uptime since the physical layer is constantly being watched. In addition, online documentation allows for a faster reaction time to problems with remote administration, and enables more productivity from those using the network.
IT groups have seen infrastructure management come full circle. From work-terminal connected to centralized computing assets to distributing computers across an enterprise"s network, the move is back toward centralization, with the data center becoming the heart of the enterprise"s business processes. Real-time infrastructure management allows true control of these increasingly critical assets.
When intelligence makes the network aware of the physical location of devices, serious problems can be resolved quickly and efficiently.For example, IP phones typically have no physical location associated with them. The switched nature of the network allows any user to operate their IP phone from just about anywhere, but it cannot identify where the call is being placed. A real-time infrastructure system manager can identify where the call is coming from because it associates the IP address of the device with a physical network port. This IP to physical layer relationship allows the voice over IP application to communicate the location of the IP phone to the emergencyresponder during a 911 call.—M.B.