Special Focus: Cabling Infrastructure

Save on data center cabling

Two new architectures offer options for low-cost, high-speed fiber-optic networks.

by Andrew Oliviero

As demand grows for bandwidth and high-speed transmission in enterprise networks, the structured cabling system in data centers is expected to maximize the efficiency of users and enable long-term performance with minimum network lifecycle costs. In this scenario, optical fiber-based network architectures can be a compelling choice for new networks and upgrades.

Traditionally, the hierarchical star topology based on copper and fiber has been used for these applications. Since the ratification of the TIA-568 standard, however, two additional network architectures have been standardized that offer users the extended reach and performance advantages of optical fiber in a cost-effective manner. These are fiber-to-the-telecom enclosure (FTTE) and centralized cabling (or FTTD).

LAN work group switches continue to be responsible for handling additional functions, such as voice over IP, video streaming, downloads and videoconferencing. There have also been significant 1-Gbps deployments to desktops and data center servers using inexpensive network interface cards, which require 10-Gbps fiber backbones. In addition, according to one report, data centers and storage area networks have been growing 30 percent annually, due to increased information generation, government data-warehousing legislation and the demand for redundancy to protect against catastrophic loss.

Meanwhile, telecom managers must support ever-increasing bandwidth and performance requirements using the existing network electronics, as long as they are still functional, and migrate smoothly to new technologies with minimum disruption and downtime.

A new standards-based solution is available that can provide higher performance to the work area, utilize existing copper electronics for the remainder of their lifecycle, offer a migration with high reliability, and simplify and reduce the cost of moves, adds and changes.

Several architectures are available to pursue these objectives. The traditional and most familiar of these is the hierarchical star. This architecture is optimized for copper performance characteristics and limitations, with a 100-meter horizontal cabling subsystem limit. It features a main cross-connect in the equipment room, with fiber backbones to remote telecommunications rooms/closets (TR/TC) and a copper horizontal.

CENTRALIZED CABLING FEATURES

More recently, the fiber-optic LAN section (FOLS) of the Telecommunications Industry Assn. created a standard to support a centralized fiber architecture for implementation of fiber to the desk (FFTD). This approach, also known as centralized cabling, features a main cross-connect in the equipment room, with fiber backbones that end at user workstations.

The third and newest approach under FOLS supports fiber to the enclosure (FTTE), with the main cross-connect in the equipment room, with fiber backbones to remote “mini-TR” (typically an enclosure). This is a standards-based commercial building structured-cabling system architecture that extends the fiber backbone from the equipment room, through the riser and the telecom room, directly to a telecom enclosure (TE) installed in a common space serving the work area. The FTTE horizontal connects the TE to the work area and may utilize copper, fiber or wireless technology.

FOLS offers a premises cost model that compares the cost of a traditional hierarchical star architecture (UTP horizontal with fiber riser backbone) to the other standards-compliant architectures: centralized cabling (FTTD), fiber-to-the-telecom enclosure (FTTE) low density, and FTTE high density. The model is designed to allow users to create a customized comparison based on the parameters of their installation.

For simple comparisons, the cost model offers several sample scenarios. The active electronics, passive cable and connectivity costs used in the sample scenarios represent aggregate costs that were derived from a public database. In all cases, the pricing used in the model represents an average price taken from at least three different manufacturers. TIA recommends that all users input their own data into the model.

The low- and high-density FTTE options trade performance for cost. In many enterprises, 32-port switches are typically deployed and configured with one 1-Gbps fiber uplink to the equipment room. This provides each workstation approximately 31 Mbps of average throughput.

HIGHEST BANDWIDTH PERFORMANCE

The FTTE low-density design offers the highest bandwidth performance to the work area because the eight-port miniswitch is non-blocking. Additional capacity exists as 200 Mbps remains on the 1-Gbps fiber backbone to the TR when all eight horizontal ports are operating at 100 Mbps. The switch is able to provide connectivity simultaneously to all eight workstations requiring 100 Mbps because the aggregate total from the eight workstations is 800 Mbps and the uplink can provide 1,000 Mbps. The high-density FTTE design represents a sacrifice in performance, but offers increased installation savings.

FTTE architectures offer the potential for significant cost savings. When low-density FTTE is deployed, the cost model estimates the total costs for network electronics, fiber and copper cable and connectivity, as well as labor and the telecom room, can result in a per-port savings of more than $198 compared to the hierarchical star.

In addition, this design offers the highest performance of any structured cabling system design. The high-density FTTE system can save even more–in excess of $272 per port, or more than 41 percent compared to the hierarchical star architecture. The TIA encourages all users to customize the cost model with their own numbers to find the most accurate estimates for their installations.

The interactive cost model is posted on the FOLS Web site at www.fols.org.

Andrew Oliviero is senior product manager for OFS, Sturbridge, Mass., and chairman of the fiber-optic LAN section of the Telecommunications Industry Association.

For more information (click here)

VERTICAL DESIGN IMPROVES MANAGEMENT

by Mark Dearing

Much attention is given to power and thermal management within data centers, but the structured cabling needs within equipment cabinets also are an important consideration. Airflow and rack utilization and organization can be impacted negatively when structured cabling needs are not part of the initial design. Airflow, in particular, is a concern, because without a means of dissipating the heat generated by the active equipment, system life and reliability suffer.

The rising cost of real estate has led IT managers to better utilize available floor space, which, in turn, means better rack space utilization. Manufacturers have responded to this need by creating products that can be mounted to cabinets outside of the traditional 19-inch space between the rails.

Vertical power-distribution units (PDU) are among the first of this type of “zero-U” equipment. Vertical PDUs are attached vertically in the back of the cabinet, either with brackets to the rail or using special button-mount equipment available with some cabinet designs. The body of the device and most of the power cords are tucked to the side of the cabinet and out of the path of airflow.

Vertical patch panels are a more recent addition to the zero-U equipment offering. These are mounted vertically in the back of the cabinet, typically on the opposite side of the cabinet from vertical PDUs to maintain separation of the power and data cables.

Like the PDUs, zero-U patch panels preserve the airflow pathway by keeping the cables and panel body to the side of the cabinet. Zero-U patch panels also can improve patch cord management by more closely aligning the patching ports with the active equipment ports, allowing standardization on shorter length patch cords for use throughout the entire cabinet. Shorter length patch cords also mean less cable to route and manage inside the cabinet.

Some zero-U panel designs include features to manage cable as it enters the cabinet and makes its way to termination at the rear of the patch panel. These panels provide channels through which cable bundles can be routed and tethered.

Of course, zero-U equipment is not always feasible. If active equipment depth is such that the cabinet rails must be positioned all the way to the back of the cabinet, little room is left between the rail and rear doors. Even if there is room for the zero-U equipment, cable bend radius is a concern. While the stranded cable used in most patch cords is more forgiving than the solid conductor cable used for horizontal runs, performance can be affected if cables are subjected to sharp bends after doors are closed.

Recessed patch panels are an alternative for applications with minimal depth in front of the panel, and in which zero-U panels are not an option. By recessing the ports three inches back from the mounting plane, these panels ensure the ability to maintain proper patch cord bend radius in close quarters. Some manufacturers also offer recessed cable management to aid in routing patch cords to the sides of the cabinet.

The trend toward higher-density cabinets can increase the potential for problems due to the corresponding increase in the number of cable drops per cabinet. Manufacturers are addressing the cabinet’s vertical cable-management needs with built-in features or optional accessories that function similar to the vertical cable management of relay racks.

One way to alleviate vertical cable-management issues and improve aesthetics is to use preterminated copper trunk assemblies. Because trunks bundle multiple cables together, the number of individual units to manage within and around the cabinet is significantly reduced.

Mark Dearing is a product manager for Leviton, Bothell, Wash.

For more information (click here)