The giants of the web are taking on the giants in the telecommunications industry when it comes to optical networking standards. Light Reading reports from the Optical Society’s annual conference today that Google and Facebook aren’t happy with the way the optical gear makers are building their wares, citing a slow timeframe for product development and unnecessary features. It’s another indication of the growing influence large web service providers are having on the computing and communications industry.
Light Reading quotes Google Senior Network Architect Bikash Koley as saying the search giant desired a more modular approach to developing optical equipment because Google doesn’t want to pay for the features the telecommunications providers want in their products. “We don’t need it; we won’t pay for it,” Koley said. Koley also explained how optics vendors need to go directly to the end customer (meaning Google) to find out what people need in the product. He says vendors rely too much on what their suppliers tell them, leading to a slower innovation cycle and products the end user finds out-of-date.
This mimics the type of influence Google and Facebook have had in the server industry with Google building its own boxes and other vendors requesting stripped-down, less-expensive boxes that run on less power. The representative from Google also echoed a similar complaint that Facebook VP of Technical Operations Jonathan Heiliger made two years ago at our Structure conference when he complained that chipmakers weren’t responding to the needs of web-scale companies and server makers and chip companies weren’t delivering the gains Facebook needed.
Facebook had a representative at the optics event as well, with Donn Lee, a senior network engineer, discussing how the industry can get together with large bandwidth users to develop multisource agreements to meet the needs of these larger players. Google’s Koley pointed to the development of a 100-gigabit Ethernet effort known as the 10×10 multisource agreement as an example. The goal of that group is to develop the lowest-cost, 100-gigabit Ethernet products for use in data center deployments.
It’s easy to paint the differences between the web services guys and the telcos as innovative companies vs the slow-moving telecom dinosaurs, but that’s not entirely fair. Telcos are slower moving, but they also have legacy systems to maintain and communicate with, as well as a mindset to deliver uninterrupted service: the hallowed “five nines.” Some of the features they require enable their fiber optic networks to integrate with their old-school copper networks, which aren’t going away anytime soon.
However, even the web companies aren’t immune to becoming tomorrow’s dinosaurs. As the web’s innovations become core to our daily lives, the companies building those innovations are building their own legacy of cheaper boxes, redundant infrastructure and more functionality being delivered via the software. There’s no reason to think that won’t change in a few decades.
Related GigaOM Pro research (sub req’d):