April 15, 2000—tax day—marked the beginning of the end of the dot-com bubble, a major driver of enterprise and IT spending. Now, as we approach tax day four years later, we are at the beginning of a new technology-upgrade cycle. After four years of avoiding upgrading their desktop equipment, IT managers are beginning to purchase new computers and the infrastructure to support them. These new computers come equipped with Gigabit Ethernet interfaces, and since users have grown to expect increasingly higher-speed connectivity, the enterprise backbone network needs a significant boost in capacity.
In addition, new regulations are expanding storage capacity requirements. Due to the Health Insurance Portability and Accountability Act of 1996, Sarbanes-Oxley, and new Securities Exchange Commission guidelines for information storage and recall, disk-based storage capacity is expected to increase at a 172% compound annual growth rate for the next several years, according to a recent report by the Enterprise Storage Group. Even organizations that are less affected by these new regulations are considering disaster recovery applications, which will fuel the need for remote SANs.
To meet these emerging needs, IT managers have begun to investigate WDM technology. Traditionally, however, WDM equipment has been very expensive, making it difficult to justify in today's fragile business climate. Every new IT infrastructure purchase must be accompanied by a clear return on investment (ROI), and the IT manager must take a good hard look at total cost of ownership (TCO).
One emerging solution is to build an inexpensive WDM system using pluggable optical modules. These modular building blocks enable flexible alternatives to costly WDM equipment for use in applications such as campus environments or between physically separated business locations.
There are three basic components to any WDM system: transponders, protection-switching equipment, and passive multiplexers, including optical add/drop multiplexers (ADMs).In traditional WDM systems, the transponder is a module that provides a conversion from multimode or singlemode fiber optics to a particular wavelength, together with some line-conditioning or performance-monitoring capabilities. In a pluggable optics design, the transponder is a flexible module, providing the electrical functions and a pair of slots for optical transceivers (see Figure 1). This module should also be highly protocol-agnostic to maintain the flexibility of the solution and provide signal regeneration and other features such as redundancy and management support. A pluggable optical interface is simply inserted into each slot on the transponder to complete the picture.
Pluggable optics—small-form-factor pluggables (SFPs) or gigabit interface converters (GBICs)—are transceivers about the size of a finger that can be adapted for many different speeds and protocols, including multimode, singlemode, copper, and "colored" interfaces for WDM. With a pluggable-based transponder, one of the pluggable interfaces is used to transmit a particular WDM wavelength. This transceiver is connected to the WDM multiplexing equipment. The other one is used to connect to the local equipment (e.g., the local Ethernet switch or router). This design separates the optics and electronics, providing cost-effectiveness, flexibility, and simplicity without sacrificing manageability.
Using a flexible SFP- or GBIC-based transponder system, particularly a protocol-independent transponder, enables any mix of Ethernet, SONET, Escon, Ficon, and video protocols to be plugged into the same infrastructure. All the pluggable modules are "hot-swappable" and can be popped out of one board and moved to another board while the system is operational.
Regardless of what nuts and bolts are used for building a system, management capabilities are important. SFP modules have built-in digital diagnostics that predict failures before they occur (via transmit and power monitoring and alarming with multiple thresholds), which equates to faster and easier equipment servicing and a far more maintainable network. This feature enables businesses with many miles between their facilities to manage the entire system from any point.
WDM comes in two flavors: coarse and dense. DWDM came first, provides dozens of wavelengths per fiber, and has been implemented by telecommunications carriers for long-haul and metropolitan transport networks. CWDM is more recent, provides up to 16 wavelengths, and has been used mostly by enterprise organizations for high-speed data transport.
Of course, with higher channel count comes higher costs in the form of more transceivers, and DWDM is around 40–50% more expensive than CWDM. However, recent announcements from component vendors indicate that the overall cost of DWDM should begin to fall dramatically. For example, at least one vendor has announced shipment of DWDM optical transceivers in the SFP form factor.
In the not too distant future, enterprise organizations will start to exhaust their CWDM wavelengths and look to DWDM to provide even more capacity. Luckily, some equipment vendors provide a migration path, allowing a network originally designed with CWDM to be upgraded to support DWDM as well. In WDM systems using pluggable optical interfaces, the transponder module that supports CWDM is exactly the same as the module that supports DWDM, providing the simplest and smoothest upgrade path possible.
The most obvious application for this pluggable WDM technology is in the campus environment (see Figure 2). Using dark fiber between the data centers of two buildings separated by several kilometers, for instance, a high-bandwidth connection can be achieved over a single fiber pair (or even over a single fiber). Multiple protocols such as Escon, Fibre Channel, Gigabit Ethernet, or even OC-48 (2.5 Gbits/sec) can be configured together much like a traditional WDM application—but in this case, much more easily and less expensively.
In another scenario, 20 or 30 channels of data connectivity may be required between a dozen buildings in a campus environment. Network requirements always tend to outgrow a fiber plant. But even with just a few available fiber strands, a pluggable-based WDM system provides a cost-effective infrastructure for networking various protocols and speeds.
The primary benefit of these pluggable optics-based WDM systems boils down to cost reduction. Pluggables provide significant ROI and the best overall TCO. There are a number of means by which that is accomplished.
Pluggable optics makes it very cost-effective to begin to build WDM installations. The up-front cost of connecting one or two wavelengths is extremely small due to the flexibility of the equipment. The system can be populated with only the number of pluggable transceivers necessary to support the
initial application. As additional wavelengths are required, more transceivers can be plugged into the equipment. Protocol flexibility is another very important benefit. A pluggable system is highly protocol-agnostic, providing nearly limitless options. Virtually any protocol can be run over the system using the same basic infrastructure.
Because of the separation between the electrical and optical devices, inventory control is dramatically simplified. For critical parts of the network, spares are a necessity—and WDM systems are rarely deployed in non-critical parts of the network. In a traditional WDM system, a spare is kept for each type of transponder module used in the network. That can be very expensive. But with pluggable optics, even with a dozen different network protocols, speeds, or distances in the network, only a few flexible spares are required. Instead of having a board for each wavelength, application, and protocol, a single SFP- or GBIC-based transponder module together with one of each color SFP or GBIC optical transceiver in use will serve as a full spare kit for any infrastructure. The small size of SFP and GBIC modules saves a tremendous amount of shelf space in addition to reducing cost.
Ongoing maintenance is a critical component of overall cost, and infrastructure management plays a key role. Pluggable optical components have internal management capabilities. They monitor power, voltage, temperature, and other required parameters. The diagnostics are also capable of generating threshold alarms. Together with a network management application that continuously monitors the network, diagnostics provide preemptive awareness of any potential problems in the optical part of the network, enabling administrators to reroute around or fix problems before they even occur.
If the network is even partially Gigabit Ethernet-based, pluggable optics enables the integration of WDM technology directly into today's GBIC- and SFP-based Ethernet switches and routers. This configuration even eliminates the transponder; a "colored" GBIC or SFP module is placed directly into a switch or router and combined with other "colors" from other GBIC or SFP modules via ADMs or multiplexers to form a complete WDM system.
Finally, it is worth keeping in mind that a single supplier that manufactures both the transponder (electrical) equipment and pluggable optics presents a huge cost advantage by eliminating mark-up costs.
With all the new data transport and storage demands being placed on their networks, IT administrators are being forced to expand their network capacity while at the same time maintaining their budget. A pluggable optics WDM solution, with its diverse deployment capability and protocol independence, provides a solution that is scalable, flexible, and cost-effective. In today's fragile business climate, these system features are coveted by anyone building a new network—or looking to extend the life of their legacy infrastructure.
Todd Rope is technology vice president at MRV Communications (Chatsworth, CA).