Optical/IP Networks

Ethernet Reaches for 100-Gig

The Ethernet camp is talking big numbers yet again, as efforts for 100-Gbit/s Ethernet and a 10-Gbit/s Ethernet PON are getting underway with standards groups.

The 100-Gbit/s effort reportedly is being organized by The Ethernet Alliance , a newly formed group that hopes to shepherd wannabe standards like this one. (See Ethernet Groups Form an Alliance.) Multiple sources say an organizing group met last week at the Institute of Electrical and Electronics Engineers Inc. (IEEE) meetings held in Boulder, Colo. -- a last-minute substitute for New Orleans, which had been slated to hold the March plenary session. The Ethernet Alliance did not return a call for comment.

The hope is to bring the matter to an IEEE call for interest in July. That's the first step down the long road to IEEE standardization, which can take two years or more. Should the plenary IEEE group green-light the idea, 100-Gbit/s Ethernet would become part of a new study group.

Ethernet jumps in factors of 10, so even though it would be cool to match the 40-Gbit/s Sonet/SDH speed grade, many Ethernet supporters are pushing for a leap to 100 Gbit/s. Some companies are talking up 100-Gbit/s Ethernet in their plans already: Extreme Networks Inc. (Nasdaq: EXTR), Force10 Networks Inc. , and Foundry Networks Inc. (Nasdaq: FDRY) say they've got backplanes designed to eventually support 100-Gbit/s Ethernet, although most of the gear in question couldn't support it right now. (See Ready for 100-Gig Ethernet? )

For now, 100-Gbit/s Ethernet is a science project with promise. Lucent Technologies Inc. (NYSE: LU) has demonstrated transmission of a 100-Gbit/s Ethernet signal, a result the company discussed at this week's OFC/NFOEC. The key point was that the experiment used 40-Gbit/s componentry. "There's no dramatically new component in there," says Martin Zirngibl, a networks research director with Lucent's Bell Labs .

The next step would be to show 100-Gbit/s Ethernet switching, something Bell Labs thinks it can do using FPGAs and tricky memory technologies. "It's a different challenge. You wouldn't put out a hero result on that at a conference like [OFC/NFOEC], but we do believe it's doable," Zirngibl says.

Looking at the more immediate future, the IEEE is ready to work on 10-Gbit/s Ethenet PON (passive optical networking), the quantum-leap successor to plain old EPON. Last week's IEEE meeting included a call for interest for the technology. The publicly available slides pitching the concept list 43 supporting companies including Broadcom Corp. (Nasdaq: BRCM), Cisco Systems Inc. (Nasdaq: CSCO), Infinera Corp. (Nasdaq: INFN), KT Corp. , Samsung Electronics Co. Ltd. (Korea: SEC), and UTStarcom Inc. (Nasdaq: UTSI).

The move is a counter to the publicity GPON has received. (See RBOCs Aim for 10M GPON Subs and Nortel, Huawei Bid on GPON.) Supporters note the possibility carriers will abandon the track of IEEE-standard PONs, turning instead to GPON, an ITU standard. GPON runs at 2.4 Gbit/s downstream and 1.2 Gbit/s upstream, outdoing today's EPON rate of 1.25 Gbit/s in each direction.

Japan, which has deployed masses of EPON, was feared to be considering GPON as its next step. Some analysts, such as Michael Howard of Infonetics Research Inc. , think that danger has subsided, thanks to talk of faster EPON grades. Teknovus Inc. , an EPON chip vendor, has been particularly outspoken about the possibilities for faster EPON. (See EPON Evangelists Talk 10-Gig.)

Of course, the long-term future could be with neither technology. An NTT Communications Corp. (NYSE: NTT) executive speaking at last week's OFC/NFOEC reportedly said he'd prefer to just jump all the way to WDM-PON, a technology that delivers one wavelength to each home. (See Novera's Got a New PON Spin.)

Korea's Electronics and Telecommunications Research Institute (ETRI) and Japan's KDDI Corp. have run trials of 10-Gbit/s Ethernet, showing the idea is viable, according to IEEE documents.

— Craig Matsumoto, Senior Editor, Light Reading

Pete Baldwin 12/5/2012 | 4:01:59 AM
re: Ethernet Reaches for 100-Gig 10 Gig EPON certainly looks like it seals the EPON franchise in Japan vs. GPON. Anybody have any further thoughts on WDM-PON, though? Will the 'next generation' (vaguely defined) come too soon to make WDM a contender?

It's been noted elsewhere on LR that Passave wasn't listed on the CFI (a doc that seemed to list everybody and their grandmother). Whassup with that?
lacitpo 12/5/2012 | 4:01:58 AM
re: Ethernet Reaches for 100-Gig Craig,

So far Passave has been biggest EPON winner in Japan ("almost" being propelled to an IPO). Does the fact that they were not on that CFI tell you something?

Passave is designing GPONs now ....

If there were a realy need for going beyond 1G or 2.5G, CWDM could come to save the day, and it could do that for both PONs.
oemarket_com 12/5/2012 | 4:01:57 AM
re: Ethernet Reaches for 100-Gig When the bitrate of a single wavelength goes to 100Gbps such kind of extremely high value, there are always questions on how economical it is compared with using multiple wavelengths and much lower bitrate each wavelength to achieve the equivalent data speed.

Many new break-throughs in components have to be achieved to realize commercially practial 100Gbps transmission.
oemarket_com 12/5/2012 | 4:01:57 AM
re: Ethernet Reaches for 100-Gig WDMs are just independent of the protocols used in the networks. It is always a low cost way to use WDM instead of increasing bit rate at each channel.
However, not sure whether EPON or GPON has left enough specifications on the laser wavelengths for ready expansion into WDM-PON.

vrparente 12/5/2012 | 4:01:56 AM
re: Ethernet Reaches for 100-Gig I don't think it's cut in stone that 10Gig EPON wouldn't use additional WDM technology. That's the purpose of moving forward -- to figure out how -- to do it.
Gnut 12/5/2012 | 4:01:56 AM
re: Ethernet Reaches for 100-Gig Marconi presenation from May 05, last few slides - the case for DWDM PONs..


ok the year is 2020 ..
Pete Baldwin 12/5/2012 | 4:01:55 AM
re: Ethernet Reaches for 100-Gig vrparente:
I don't think it's cut in stone that 10Gig EPON wouldn't use additional WDM technology. That's the purpose of moving forward -- to figure out how -- to do it.

True -- one of the options for 10GEPON is to send four channels of 2.5 Gbit/s each. It's a possibility.

Given that 10GE will be reasonably well established by the time the 10GEPON standard gets done, I'd rather see them go with straight 10Gig. But i can see them keeping the 4x2.5 option in the standard -- could be useful for early implementations.
nwave 12/5/2012 | 4:01:52 AM
re: Ethernet Reaches for 100-Gig I don't think it's cut in stone that 10Gig EPON wouldn't use additional WDM technology. That's the purpose of moving forward -- to figure out how -- to do it.

Agreed! The question is how the 10GPON and the 2.5GPS folks can present a clear migration path into WDM. WDM PON costs at the moment currently seems like a $$tretch- Five years from now this may not be the case as the WDM PON market matures and more vendors move into the space. The question is what technology 2.5GPON or 10GEPON or perhaps both can and will build a bridge towards WDM PON.

The IEEE voting to move 10GEPON technology forward is the right move now as a valid response to the bandwidth requirements quickly moving into the network. These BW access requirements are growing by leaps and bounds, very far beyond the initial expectations of both the original GPON and EPON committees (which both saw inception in the early 2000GÇÖs) bandwidth expectations. With HDTV ready devices (production and reception) pod-casting, video RSS feeds, video-podcasting and IPTV based sliver-casting and many bandwidth intensive services moving rapidly into the network the last mile bandwidth question becomes very pertinent. On top of that Super and Ultra HDTV technologies are in the labs now and are likely to emerge very rapidly into consumer based products. The bandwidth promise of 2.5GPS that is eventually promised by the GPON standard vendors may likely end up "to little bandwidth but to late to market". If current projections are correct mature 2.5GPS GPON systems may not reach a stable, mature and network deployable product until early or even mid 2008. Based on historical trends of Ethernet and GÇ£Ethernet economicsGÇ¥ I would expect initial 10GEPON systems start to emerge pretty soon after this date. The key question then of what to deploy widely in the network becomes a very pressing and very pertinent question. P2P fiber is an alternative but costs are also very prohibitive. A quick and very inexpensive BW expansion can be offered now in current deployed GEPON systems by lowering the split counts to 1:16, 1:8 or even 1:4, using a fashion similar to what that CATV broadband providers are currently partitioning their networks to respond to increased DSL speeds, to counter competitors. Many GEPON providers are starting to do the same thing effectively and cost effectively to meet very high BW customers requirements.

The issue at the moment that I see for WDM based PON is lack of vendors, with Novera Optics as a current exception (It would be interesting to know of other vendors in this market). Although this market will probably mature five to six years out as more vendors and cost effective product really begins to emerge. The current market costs for WDM are highly prohibitive. The following paper presents an outline of how this bridge from TDM merged with WDM may be accomplished at a reasonable cost.

Sign In