BALTIMORE, MD--A development that could help carriers slash costs and move towards making their backbones future-proof was unveiled today (March 7) at the Optical Fiber Communication (OFC) exhibition in Baltimore.

Xros Inc. http://www.xros.com (pronounced kai-ros), a Californian startup, unveiled an optical cross connect called the X-1000 that breaks new ground in two respects.

First, it can scale to 1,152 pairs of input and output ports – a big advance on existing developments and the first cross connect to reach the 1,000 port threshold often cited as a minimum requirement by carriers. They say they need optical cross connects of this size to connect together the massive increase in the numbers of wavelengths in their backbones – a result of the widespread deployment of dense wave division multiplexing (DWDM) technology.

Second, the X-1000 is all-optical, which makes it future-proof. Light isn’t converted into electrical signals to be switched, as is the case with some optical cross connects. This means the X-1000 will carry whatever can be contained within a wavelength – now and in the future. It won’t need upgrading to carry higher bandwidths as new transmission technologies arrive.

And unlike some other all-optical cross-connect, the X-1000 also is a genuine switch rather than an automated patch panel. Connections between input and output ports are set up automatically, in response to signaling, in less than 50 milliseconds. That’s fast enough for protection systems – setting up alternative paths across backbones in the event of a line or equipment failure.

Scalability SecretLike other vendors with reasonably large scale all-optical cross connects, Xros uses micro-electro-mechanical system (MEMS) technology – arrays of tiny mirrors – to bounce light from input to output port in the X-1000.

Xros has managed to squash 1,152 mirrors, each measuring 2 millimeters square, onto a six-inch square piece of silicon. Each mirror is coated with gold leaf to reflect the maximum amount of light (98 percent according to Reznick) and is hung in two concentric frames, in the same way as a compass is mounted on gimbals, so it can be tilted in any direction by applying electrical current.

The real secret of Xros’s scalability comes from being able to measure and thus control the position of the mirrors extremely accurately (within one five millionth of a degree, according to Xros), by measuring the strain in pivots.

In the X-1000, two of these arrays of mirrors are placed opposite each other, and arranged so that light enters the switch, hits a mirror at about 45 degrees, is bounced to a mirror in the opposite array, which then bounces it out of the switch. Any mirror can bounce light onto any other mirror in the opposite array.

Other vendors, such as Siemens AG http://www.siemens.com, are using less sophisticated MEMS developments. In Siemens’ case, a single matrix of mirrors is used -- which flap up to deflect light, or stay down to let it pass.

Such architectures won’t scale, according to Greg Reznick, Xros’s president and CEO. A 1,000 port cross connect based on this technology would need 234,000 mirrors, he says. Building such a huge, complex device isn’t practical. Moreover, the path length of the light varies, raising further complications, he contends.

So far, Siemens has only demonstrated a 32 port optical cross connect (see Siemens Launches Optical Cross Connect “Solution”). Lucent Technologies http://www.lucent.com has also demonstrated a 256 port optical cross connect based on yet another variant of MEMS technology. Both vendors say they hope to develop 512 port models by the end of the year.

It’s important to note that Lucent and Siemens are developing what amounts to automated patch panels – devices that have to be configured manually, albeit from a remote console. Xros’s X-1000, on the other hand, automatically switches light waves in a matter of 50 milliseconds, as already noted.

Competition is ElectricXros’s toughest competition, in fact, is likely to come from vendors of optical cross connects that have electrical cores – notably Lightera (acquired by Ciena Corp http://www.ciena.com), Monterey (acquired by Cisco Systems, Inc. http://www.cisco.com, Tellium Inc. http://www.tellium.com and Sycamore Networks, Inc. http://www.sycamorenetworks.com.

All of these vendors are developing equipment that automatically set up and tear down wavelengths in response to signaling, just as Xros is doing. And although none of these vendors currently have 1,000 port cross-connects, it’s only a matter of time before new switching fabrics arrive and allow them to beef up their current developments. Further, at least one new start-up is working on a cross connect that will scale to many thousands of ports (Brightlink Networks Inc.).

However, it’s important to bear in mind that optical cross connects with electrical cores won’t be as future proof as their all-optical equivalents. Vendors making a cross connect handling 64 OC48 channels, for instance, will upgrade the device to support OC192 bandwidths by combining four OC48 streams, so the device ends up with only 16 ports. If they upgrade again to OC768, then the cross connect ends up with only four ports. “What looked like a system is now a component,” says Reznick. In contrast, an all-optical cross connect always keep the same number of ports and doesn’t need upgrading at all to handle higher bandwidths.

Right now, the X-1000 represents an optical cross connect offering “OC192 at OC48 prices,” says Reznick. As bandwidths increase, the price advantage will grow, he adds.

Still, Xros’s X-1000 still has some snags. For a kickoff, light becomes weaker as it passes through its equipment, and this might mean that the X-1000 can’t be used to replace existing electrical cross-connects in some circumstances. The adjoining DWDM terminals might not have powerful enough lasers to pump light across to the X-1000, according to Reznick.

And although the X-1000 can automatically reroute wavelengths around failures, there’s no mechanism to sense failures at present. A signaling standard is needed to carry this information from the DWDM terminals to the cross-connect, says Reznick,

It’s also early days for Xros. Its technology is unproven. It’s showing a prototype at OFC , expects to start trials this summer and begin commercial shipments early next year. And of course, the startup can’t provide the support that many carriers would expect. But if its technology lives up to expectations, Xros is bound to be bought before too long.

-- by Peter Heywood, international editor, Light Reading http://www.lightreading.com