It has been a busy few weeks since my last blog, and the next several months are looking just as busy. I just attended the Ethernet Technology Summit last week, and next weekend I leave for OFC. The week after I get to visit Singapore for the IEEE 802 Plenary. Lots of activities underway related to 100GbE and looking beyond. Just some quick thoughts on all of this:
• Ethernet Technology Summit – great conference this year with a lot of interesting presentations. I attended a workshop on beyond 100GbE and the interest level was high on the next rate of Ethernet. Daniel Blumenthal of the Terabit Optical Ethernet Center (TOEC) gave an interesting overview of the group, which then led into a compelling discussion on the technical challenges that will need to be overcome. Optics was obviously an integral part of the talk, but I had to keep reminding everyone that we haven’t left the electrical world yet, and people better be considering how they are going to get all the data from their ASICs to the optical module. 25G is the next horizon for electrical signaling. So we are looking at a x16 interface for 400GbE or a x40 interface for Terabit Ethernet. The next day I attended a panel on the state of 40GbE and 100GbE. Once again the role that 4x25G interfaces will play in cost reducing 100GbE was a featured topic of discussion.
• OFC 2011 – I am really excited by the prospect of this year’s show. At last year’s show, with the impending ratification of the 100GbE standard, we saw optical vendors showing their CFP modules targeting the 100GBASE-LR4 specification for 10km. A whole year later, what will the optical module vendors showcase? Personally, I am looking forward to seeing if there are any developments regarding the optical nature of the CFP module. Vendors have had a whole year to move past the 1st generation modules, based on discrete components, to drive cost and power down. This will be key to adoption and deployment of 100GBASE-LR4, as well as potentially addressing the issue of whether a new optical specification is needed for 2km.
Also, it will be interesting to see if any announcements are made by member companies of the CFP MSA regarding the roadmap that has been publicized over the past few months regardingrelative to the future higher density CFP2 and CFP4 modules. It is clear that there is industry momentum behind the development of 25G+ electrical signaling that will play a critical role in the cost reduction of 100GbE, as well as driving the port densities that will be needed by systems both inside and outside of the box. Consider the work in the OIF on its CEI-25/28G efforts, as well as the new 100Gb/s Backplane and Copper Cable Study Group in the IEEE 802.3.
• Speaking of the 100Gb/s Backplane and Copper Cable Study Group, its next meeting in Singapore should prove to be interesting, as it is clear that the technical challenges in driving 4 lanes of 25Gb/s across a backplane will be daunting. Hard to believe that just 10 years ago I was working in the 10GbE project on driving 4 lanes of 3.125 Gb/s. Sign….I know it’s progress… but man it does make me feel old sometimes.
The Ethernet Bandwidth Accessment Ad Hoc will also be meeting in Singapore. Industry discussions regarding the next speed of Ethernet are continuing, and I am seeing further evidence of the prediction of the need for Terabit Ethernet.
It is clear that the industry has a lot of work in front of it!
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.