& cplSiteName &

ImmenStar Joins EPON Race

Craig Matsumoto
5/30/2006
50%
50%

An Ethernet PON chip startup will launch itself into the market this week, with chips designed by a Force10 Networks Inc. engineer and bold claims about a customer win in Japan.

ImmenStar Inc. is readying a May 31 press conference to introduce itself, playing up the fact that the company has spent two years in stealth mode. The element of surprise is kind of blown, though, as ImmenStar had a previous launch -- on May 18, at the NetEvents Asia Press Summit in Singapore -- and has posted product descriptions to the Web.

The EPON market already seems to have its fill of chips, with competitors including Centillium Communications Inc. ; Passave Inc. , recently acquired by PMC-Sierra Inc. (Nasdaq: PMCS); and Teknovus Inc. And they aren't standing still; Teknovus, in particular, is preparing for the next generation of EPON. (See PMC Preps PON Plans and EPON Evangelists Talk 10-Gig.)

"We're a little bit late, as you can tell," says John Wuu, ImmenStar's CEO. "We watched the Japanese market, and PON started picking up very well. We believed we could do something different."

ImmenStar so far has been on the fast track to catch up. The company got chips produced in one year and claims to have landed a multimillion-dollar order from Japan, with delivery due next month.

ImmenStar isn't saying who the customer is, but one source pegs it as UTStarcom Inc. (Nasdaq: UTSI) -- which, not coincidentally, invested $2 million in ImmenStar, according to Securities and Exchange Commission (SEC) filings. (ImmenStar has raised one round of funding, led by Pac-Link Management Corp. , for an undisclosed amount. Wuu notes that UTStarcom owns a "minor" percentage of ImmenStar.)

UTStarcom sells equipment to Softbank BB Corp. for the Yahoo Broadband service in Japan. UTStarcom had been using EPON chips from Passave Inc. , but that relationship allegedly soured when the Passave chips didn't work properly, according to a lawsuit UTStarcom filed last year. (See UTStarcom Sues Passave.)

Passave even considered buying ImmenStar at one point, Wuu acknowledges. "That was a long time ago. They never talked to us directly. They hinted through a different party," he says, noting that he wasn't involved in that conversation himself. Passave officials declined to comment on the whole situation.

Naturally, other competitors are wise to ImmenStar, too. "They're serious, and they're smart guys -- but about two years behind us," says Lowell Lamb, vice president of marketing for Teknovus. "They're a competitor. It's a sign that the market's growing up."

Trying to stand out, ImmenStar packs its chips with features such as quality of service (QOS) with video in mind, and the company claims its chips let carriers guarantee the bandwidth dictated in service-level agreements.

But the most obvious difference in its parts is that the optical line terminal (OLT) chips have four ports, apparently the first of their kind to do so. It's a step that was probably inevitable in PON chips, and it could lead to more densely packed OLTs for central offices. Having four ports also creates a built-in redundancy, as traffic can be shifted to a different port if something goes wrong.

ImmenStar's founders include Eugene Lee, an ASIC designer who was chief architect and employee no. 1 at Force10, the high-end Ethernet switch vendor -- and he's brought three other Force10 employees in his wake. "A lot of people in PON chips have broadband knowledge. They don't have the router/switch domain knowledge," Lee says.

Other key execs come from systems companies as well. CTO Charles Chen hails from Mahi Networks, which got absorbed into Meriton Networks Inc. (See Meriton Buys Mahi Networks.) Wuu is a Bell Labs veteran who was also Telcordia Technologies Inc. managing director for Asia/Pacific and a sales VP at IP Unity Corp. , arriving there when the company acquired EmpowerTel.

Lee developed Force10's first two generations of chips, but because the company sells to such high-end networks, he found himself a bit stranded. "I found out that probably for the next five years, I don't need to do anything," he says. That, plus a discovery that "the edge of the network is more fun, with more challenges," led him to leave Force10 to start ImmenStar.

Lee notes he's held onto his Force10 stock -- and no, he doesn't know when that company will go public either. (See Force10 Revs Revenues.)

Most of ImmenStar's 30 employees work from Santa Clara, Calif., with a handful in Shanghai and one staffer starting up a customer-service office in Taiwan.

— Craig Matsumoto, Senior Editor, Light Reading

(9)  | 
Comment  | 
Print  | 
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
Pete Baldwin
50%
50%
Pete Baldwin,
User Rank: Light Beer
12/5/2012 | 3:52:54 AM
re: ImmenStar Joins EPON Race
The quad OLT ... I like the idea, but I've also heard a hint from one competitor that there's no pull for such a thing yet. What do you folks think?

I also like that Immenstar has roots in Force10. But I'll feel more comfortable about them once they get a customer beyond UTStarcom.
++++++
50%
50%
++++++,
User Rank: Light Beer
12/5/2012 | 3:52:53 AM
re: ImmenStar Joins EPON Race
I like the Quad OLT idea too. Why not? 4+ chip suppliers in the market will drive EPON component costs down, and force more rapid innovation (ex. 10G-EPON, 8-port OLT, 16-port OLT, etc).

While I'm not surprised UTStarcom would drop Passave/PMC-Sierra (bugs/design issues), didn't UTStarcom invest in Immenstar; if so, their one-customer win is less impressive.
Pete Baldwin
50%
50%
Pete Baldwin,
User Rank: Light Beer
12/5/2012 | 3:52:52 AM
re: ImmenStar Joins EPON Race
Ah, Teknovus's got one. My bad for not remembering that.

Something else, too:
>I am not sure that the costs of PON systems is very heavily dependent on the cost of OLT silicon. One could easily fit 4 single PON controllers on a line card as well.<

Good point -- moreover, are carriers all that anxious for higher density PON yet? I guess since we're talking about EPON and the Asian market, the answer would be yes. North America is probably content with whatever's cheaper, be it single or new-generation quad (although Fujitsu is going for big density with GPON:
http://www.lightreading.com/do...
paolo.franzoi
50%
50%
paolo.franzoi,
User Rank: Light Sabre
12/5/2012 | 3:52:52 AM
re: ImmenStar Joins EPON Race

http://www.teknovus.com/tk3722...

Already a quad on the market. I am not sure that the costs of PON systems is very heavily dependent on the cost of OLT silicon. One could easily fit 4 single PON controllers on a line card as well.

seven
moonlighter
50%
50%
moonlighter,
User Rank: Light Beer
12/5/2012 | 3:52:51 AM
re: ImmenStar Joins EPON Race
right, from the link shows Teknovus has a quad OLT ports but has one aggregated uplink 1 ge port compared to Immenstar's quad epon ports with 4 ge line rate uplink ports from the their website. Immenstar has their technical advantage with good features even though they enter the market a bit late than their competitors.

moonlighter.
paolo.franzoi
50%
50%
paolo.franzoi,
User Rank: Light Sabre
12/5/2012 | 3:52:50 AM
re: ImmenStar Joins EPON Race

I am not sure that is a big deal moonlighter. The OLTs themselves have 1 - 4 GigE uplinks for 64 - 128 EPONs.

On the other hand, they have UTStarcomm as a backer and customer. If UT can get YahooBB really going, it would be good business.

The rest of it goes primarily to Passave and secondarily to Teknovus.

seven
materialgirl
50%
50%
materialgirl,
User Rank: Light Beer
12/5/2012 | 3:52:48 AM
re: ImmenStar Joins EPON Race
Dear Seven:
Big IF on that Yahoo BB network in Japan. UTSI has shipped gobs of equipment for that network over the past year and has barely been paid. UTSI seems all talk and zero follow-through.
Mark Sebastyn
50%
50%
Mark Sebastyn,
User Rank: Light Beer
12/5/2012 | 3:52:41 AM
re: ImmenStar Joins EPON Race
I think the advantages of having a Quad OLT arise from the ability to more efficiently do statistical multiplexing onto the uplink side. If you have four discrete OLT chips the indivisual GE uplinks from each device need to be fed into an L2 aggregation chip that will make drop decisions. Having four OLTGÇÖs on one chip let you merge the control plane and make better decisions on DBA and discards as you forward. Theoretically, you would get better QoS and utilization.

BUT - as of right now, fairness on the upload side isnGÇÖt that important. Few applications exist that push the envelope in the upload direction - Enterprise applications and circuit emulation are two that would, but neither have been deployed over PONGÇÖs in any volume.
paolo.franzoi
50%
50%
paolo.franzoi,
User Rank: Light Sabre
12/5/2012 | 3:52:40 AM
re: ImmenStar Joins EPON Race

You would get more effiency gain by eliminating 8B10B encoding and going with scrambling.

seven
Featured Video
Flash Poll
Upcoming Live Events
March 12-14, 2019, Denver, Colorado
April 2, 2019, New York, New York
April 8, 2019, Las Vegas, Nevada
May 6, 2019, Denver, Colorado
May 6-8, 2019, Denver, Colorado
May 21, 2019, Nice, France
September 17-19, 2019, Dallas, Texas
October 1, 2019, New Orleans, Louisiana
October 10, 2019, New York, New York
November 5, 2019, London, England
December 3, 2019, New York, New York
December 5-3, 2019, Viena, Austria
All Upcoming Live Events