Researchers at the University of California, Santa Barbara (UCSB) are working on the next evolution of Ethernet – Terabit Ethernet. UCSB Professor of Electrical and Computer Engineering Dan Blumenthal told LightReading that the goal of the recently created Terabit Optical Ethernet Center (TOEC), is to create Terabit Ethernet (TbE) which runs at 1 trillion bits per second by 2015 and to follow it up with 100Tbit/s Ethernet by 2020.
Professor Blumenthal explained to LightReading that he wants the TOEC and its partners to produce something the industry can use, not a one-time lab experiment that only works with duct tape and glue. “We’re not talking about lab hero experiments,” Blumenthal told LightReading. The real-world focus of TOEC has helped attract partners like Agilent Technologies Inc. (NYSE: A), Google (NASDAQ: GOOG), Intel Corp. (NASDAQ: INTC), Rockwell Collins Inc., and Verizon Communications Inc. (NYSE: VZ) to help with the research. I wrote about Intel’s TBPS efforts back in July.
Terabit Ethernet is hard
TOEC could probably use the help because developing TbE is looking like no simple task according to LightReading. Bob Metcalfe, Ethernet’s creator, and now a Polaris Venture Partners partner, speculated two years ago that a terabit standard might need a rethinking of everything, even the fiber itself.
Based on current UCSB research, professor Blumenthal speculates that TbE may include:
- Photonic integrated circuits (PICs) are a must.
- Coherent receivers, but at a scale well beyond what’s being used for 100Gbit/s Ethernet. A likely candidate is 1,024-QAM: quadrature amplitude modulation (QAM) transmitting 10 bits per symbol, a scheme likely to require 100GHz electronics.
- To make that coherent receiver energy-efficient, TOEC is “trying to move a lot of what’s in the digital signal processor into the optics,” Blumenthal says.
- New materials for fiber-optics aren’t out of the question. “We won’t start out with that, but it’ll move in that direction,” Blumenthal says.
- Other items on the TOEC shopping list include optical phase-locked loops, new semiconductor optical amplifiers (SOAs), and methods for drastically lowering on-chip optical losses.
The questions go beyond the optical layer. To make operations more synchronous padding and frame delineation were added to 10Gbit/s and 100Gbit/s Ethernet, Blumenthal pointed out. “Do we keep doing that? Or do we go purely asynchronous? We don’t know yet. …Once you put the word ‘Ethernet’ in there, it’s not about just transmission. It’s about being backward-compatible. That’s the beauty of Ethernet. We can’t lose that essence.”
rb-
The need for TbE is real (I first wrote about Intel’s TbE efforts here) and being driven by video. More video is already riding over existing networks. “We’re going to need much faster networking to handle the explosion in Internet traffic and support new large-scale applications like cloud computing,” Professor Blumenthal told Physorg. Stuart Elby, Vice President of Network Architecture for Verizon told Physorg, “Based on current traffic growth, it’s clear that 1 Terabit per second trunks will be needed in the near future.”
Facebook is already looking at TbE in their data centers. PCWorld reports that at the Ethernet Alliance‘s Technology Exploration Forum, Donn Lee, a Facebook Engineer said, “… there is already a need for 1 terabit.” Facebook has so many servers, and those servers can process data so fast, that they could fill 64 Terabit Ethernet pipes in the backbone of one data center, Lee said.
Related articles
- Terabit trial gives Telstra some backbone (go.theregister.com)
Ralph Bach has been in IT long enough to know better and has blogged from his Bach Seat about IT, careers, and anything else that catches his attention since 2005. You can follow him on LinkedIn, Facebook, and Twitter. Email the Bach Seat here.