Why Optical Crossconnects
Are Photons Really Better Than Electrons?
Optical Crossconnects (OXCs) can be constructed with either an electrical switching fabric or an optical one, which we call Photonic Crossconnects (PXCs). At Glimmerglass, as a developer and manufacturer of Photonic Crossconnects, we frequently get asked the question “Why photonic?” We, of course, think the question should be “Why OEO?” as in “Why would anyone want an OEO crossconnect?” The history of the telecommunications industry over the last couple of decades has been, to a large extent, a history of the increasing penetration of optics. Ever since the first practical optical fibers and lasers got a foothold, much of the progress has been measured by the elimination of electronics from the network.
Erbium Doped Fiber Amplifiers (EDFAs) were a major milestone not only because they ushered in the era of Dense Wavelength Division Multiplexing (DWDM) increasing fiber efficiency, but they also eliminated costly, inflexible OEO regeneration. Would DWDM have been any way near as important an innovation if each and every wavelength had to be terminated and regenerated at every hut?
The advent of fixed Optical Add Drop Multiplexers (OADMs) further reduced the number of OEO conversions required in the network by permitting through traffic to remain in the optical domain. Only those wavelengths destined for a given site or needing grooming at that site needed to be dropped and converted to electrons for processing. Reconfigurable Optical Add Drop Multiplexers (ROADMs) are the latest in this line of evolution. While not further reducing the number of OEO conversions, ROADMs do significantly ease the operational burden. The current migration to ROADMs points out that there are benefits to eliminating OEO conversions beyond just eliminating the costs of the laser transmitters and receivers.
As ROADMs start to evolve to take on some of the functionality of crossconnects, some electrically, some optically, let’s look at a few of the benefits of Photonic Crossconnects.
Lower costs – It is true that the costs of 10GE transmitters and receivers have been dropping rapidly. Furthermore, if history is any guide, we can anticipate further reductions in the future. This is nothing new. Ten years ago it was the costs of OC-48 transmitters and receivers which were rapidly falling. That didn’t stop the increased penetration of optics and neither will this. At some point, a new higher performance technology emerges with higher costs, but lower cost per bit, witness AT&T’s recent announcement of the introduction of 40G wavelengths into their backbone.
In addition, we shouldn’t restrict our focus to transmitter/receiver costs. A lot of high speed electronics is required to build a large crossconnect. A 160x160 non-blocking crossconnect with 40G line cards must have an interconnect network supporting an aggregate bandwidth of 6.4Tb/s. This is not cheap. Furthermore, it is important to remember that networks are always in a state of flux. The cost of upgrading an OEO 10Gb/s crossconnect to 40Gb/s is dramatically higher than the cost for upgrading a Photonic Crossconnect.
Reduced space, heat and power – PXCs will always be orders of magnitude better than OEO crossconnects in space, heat and power. Today a 160x160 non-blocking PXC is available which fits in 4U (7”) of rack space and consumes only 80W of power. Even larger switches are on the way. The corresponding OEO switch with 10G or 40G interfaces would require several full racks and several kilowatts of power. The potential elimination of large HVAC upgrade and operational costs alone is often an important justification for Photonic Crossconnects.
Transparency – Bitrate and format transparency has always been an important benefit of PXCs. No network operator wants to face the Big Bang approach of upgrading all facets of their network to a newer, better, faster technology. All upgrades must be incremental as justified by traffic. As in the case of ROADMs, Photonic Crossconnects permit network operators to easily mix different traffic types and upgrade without forklifts.
Quality Of Service (QOS) – This is a relative newcomer to the list of significant benefits of PXCs. While always a PXC feature, the rapid rise in streaming media and integrated IP networks has caused a new focus on methods to transform best-effort IP networks into the efficient guaranteed-delivery networks required. Photonic Crossconnects are not only non-blocking, but have virtually no arrival time jitter or packet loss.
There are still portions of telecommunications networks where optics
is not the solution, for example, where grooming and wavelength conversion
are required. But for large crossconnects photons, not electrons, are
the most effective solution.
