Could optical networking be coming to a data centre near you?

28 Feb 2012

Cisco’s $271 million acquisition of Lightwire represents a hefty bet that silicon photonics has a future in mainstream corporate IT, as Jessica Twentyman explains. 

Cisco now aims to put Lightwire's technology in the hands of data centre managers, a group that's increasingly forced to accommodate huge volumes of traffic between servers and storage, triggered by trends such as big data, mobile technologies and virtualisation, at a time when they’re also working against a backdrop of high energy costs and IT budget constraints.

Right now, optical networking technology – in which data is conveyed via pulses of light rather than pulses of electricity – isn’t generally seen in corporate IT set-ups. Typical customers tended to be telcos and other carriers who use it to beef up metro, regional and long-haul networks and do not shy away from extremely high prices to achieve their goals.

But Lightwire’s main attraction for Cisco is that it has pioneered a proprietary process for making high-speed optical transceivers, a type of chip that can transmit data along a precise beam of light. That process uses traditional CMOS [complementary metal-oxide semiconductor] processing and it’s a far cheaper way to build optical chips than previous approaches.

This means that these chips can be priced for mass use and incorporated into the switches, routers and other kit that carry data around mainstream enterprises, as well as in more specialist carrier networks.

“The acquisition of Lightwire will support our data centre and service provider customers as they manage the continuing deluge of network traffic alongside tight capital and operating budgets, said Surya Panditi, senior vice president of Cisco’s service provider networking group, in an official statement. “With the combined know-how from Cisco in silicon design and Lightwire in CMOS photonics, we will transform Cisco’s optical connectivity business to an integrated technology platform that supports our customers’ burgeoning need for cost-effective high-speed networks.”

In buying Lightwire, Cisco builds on its 2010 acquisition of another optical networking specialist, CoreOptics. It also gets some of the finest minds working in chip-based optical networking – or silicon photonics – today. And Ameesh Divatia, Lightwire’s president and CEO, was Cisco’s chief product strategist for optical transport products between 1998 and 2001, which may help to smooth the integration of the two companies. 

The purchase also marks a return to the M&A trail for Cisco, which has traditionally been one of the IT industry’s most voracious shoppers – a habit it was forced to curb during most of 2011 while it underwent company-wide restructuring. Earlier in February 2012, however, on its second-quarter earnings call, Cisco CEO John Chambers told investors, “we expect to be more active with acquisitions in the quarters to come.”

The Lightwire deal is a clear bet on the growing importance of silicon photonics, say industry watchers, and an explicit challenge to competitors in this market, which include IBM and Intel, plus a number of smaller firms including Luxtera, Kotura and Chiral Photonics. All are US-based but the technologies they’ve developed are still some way off from being fully commercialised.

That won’t happen until 2016, according to a market study by Marketsandmarkets.com - but silicon photonics is still seen as a ‘hot’ area and the same report forecasts the market to grow from $1.12 million in 2010 to $2.02 billion in 2015, a compound annual growth rate of 78 percent. Just hours after news of the Cisco/Lightwire deal went public, Luxtera announced it had closed a $21.7 million round of funding.

The race is now on to create viable networking infrastructures for what many refer to as ‘exascale computing’ – computing capabilities that outstrip today’s petascale limits by a factor of one thousand. Today’s copper won’t be able to cope. Data movements in this exascale era will need to use light not electronics. By converting electronic signals into light for transportation right down at the chip level, silicon photonics seems to propose the most likely solution so far.