1996: Ipsilon - Red Herring Article

Red Herring magazine article about Ipsilon Networks. Featuring a bald Tom Lyon.

Transcript

1. 1 < , -.. LLLLLLLLLLLLLLLLLLLLLLLLL - M“ ~t P R O F I L E IP switching has the power of ATM and the popularity of the Internet. ‘E networkings biggest problems—the ..-3 5‘,-, ,, » - A ~ bandwidth crunch. I Speeding up the pipeline - - I Virtual connections required in networks G are growing geometrically with the in- creasing density ofservers and database en- gines; graphics and multimedia content are putting more demand on performance; and the pure number of users have increased " looking at Tom Lyon, the founder (see sidebar“Channeling the new medium,” network trafc—so bandwidth problems and chief technical ofcer of Ipsilon p.50). Lyon got enough encouragement to loom large. And as the Internet grows, so Networks, you know he has a pas- keep going and Ipsilon Networks was does the intranet, as more and more private k _. ble the erformance of our soft- ;, ’ . . and Andy Rachleff of Merrill the challenge. Traditional network routers P at ware by lune Ist,” says Mr. Lyon. I, Pickard led the rst round invest- introduce delay and today’s LAN switches sion for his work.“Back in Ianuary, ti‘ ”l"‘_l:, 11, born. enterprise networks are becoming based on I challenged our engineers to dou- SW P I i Ion Feiber of Mohr Davidow lP. But today’s networks aren’t always up to “My deal was that ifthey shaved the ment of $3.5 million in Ipsilon in don’t always route. So how to improve this overhead, I’d shave my head. But why I November 94. Lyon had known Feiber situation without a costly revamping ofthe go to such extremes? It all comes down to for 15 years, and worked with him when entire network?“Ipsilon’s architectural goal the big idea. Feiber was hired as employee number 35 at is to make IP faster without changing it,” ' The of both Lyon was acquainted with Rachleff says CEO NeSmith.“\A/e’ve designed a so- ecause ofa Merrill Pickard investment in lution that ts cleanly into existing IP net- Lyon’s big idea is to get Internet Protocol brother Bob Lyon’s startup, Legato Systems. work architectures and works with current (IP) to go faster by harnessing ATM’s raw But as much as anything,the investors were IP applications and network management switching power. drawn by the idea.“Lyon was trying to cut tools.” With IP switching, Ipsilon is taking Lyon was employee number8at Sun Mi- today’s networking Gordian Knot," says IP and layering it on top of high-perfor- crosystems where, for 12 years, he says, “I mance ATM switching fabrics.It’s designed was a member of the networking mists TOQ €‘lZl'1 8 P TI l'1 8)’ would try ‘CO to combine the intelligence and control of group who nevertheless successfully es- IP routing with the high speed and capac- caped a career as a UNIX expert.” Lyon also S O LV 6 O F1 6 Of H etwO I‘ l("| n g '5 ity ofATM switching. Val Sribar, analyst at joined the ATM Forum at its formation in the META Group, calls this “a refreshingly Forum, it was only too clear that a standards Ipsilon claims its agship product,the IP bodywasn’t going to come up with anycre- ba ndwidth CPU |'1Ch . Switch ATM I600,supports aswitching ca- ative solutions. In ’94, Lyon left Sun “be- pacity of up to 5.3 million IP packets-per- cause I couldn’t gure out how year nu1n- Feiber. “The problem of implementing second (PPS) and port throughputs of 155 ber 13 would be any different.” He became ATM on top ofswitching without piling up Mbps. It’s priced at approximately $3,000 an entrepreneur-in-residence at the venture more iron.” per port. An alternative is gigabit Ethernet capital rms of Mohr, Davidow Ventures, Feiber and Rachleffhelped Lyon recruit switching, which according to LeBaron, and Merrill, Pickard, Anderson & Eyre. an experienced management team that “basicallyinvolves cranking Ethernet up to Here he incubated the big idea that was includes president and CEO Brian Ne- another order of magnitude.” Yet she says turned down bythe Forum~combining IP Smith, formerly of Newbridge Networks, this is only good for conventional data, not packets with ATM switching. The result and vice president of marketing Larry telephony, qualityvideo,or other multime- was IP switching. On August 31, ’94 in Bill Blair, co-founder of Kalpana. Ipsilon has dia components. IP switching seems more Davidow’s ofce, Lyon presented his idea to also put together what Melinda LeBaron, promising. Hermann Hauser, chairman of ATM Ltd. network analyst with the Gartner Group, What are the market prospects for the I * - describes as “a stellar en ineerin team.” “* xi ‘ Together they would try to solve one of Q3, -_;\1,;§{<,‘,“,“"t\'y’\l=A‘v1 -‘—.:1,,; t as P ll,“ g \ 1 ‘ ‘ \ 1 " it ' *»~,,,t;:»- '2 ye — “ “'4 1 < ~ A’ rt <'~'§’="l é‘ 1 l l l I 1 <_> </1 m >-< >-< to J THE~l‘,HER?ING AUGUSIl996 1 l l 1 l I991. After three years of debating at the big Q 8 St p PO b L8 ms——t h 6 innovative approach.” l l l l l l l l
2. L WT > I l l IP friendly: Lyons d~ Co. are revving the worlds most popular protocol. switching technology to switch vendors (ATM initially, but not exclusively) to cre- ate the switches. It already has partners on the desktop (ATM Ltd.) and the WAN backbone (Hitachi). lpsilon's own systems business will focus on the corporate back- bone. Other important partners include Digital Equipment Corporation (DEC) and Integrated Device Technology (IDT), both ofwhich invested in Ipsilon as part of a $22 million third round f1nancing.Addi- tional technology partners are MMC Net- works, Scorpio Communications, and Radcom. LeBar0n believes this broad part- nering strategy puts pressure on network- ing’s Big Three—Cisco, 3Com, and Bay Networks—to eventually port the Ipsilon switch, or one like it. She puts this effort on a rough timetable. “If it is to succeed, Ip- silon’s strategy will need to show denite signs of success by the end ofcalendar ’97," IP switching solution? Tom Erickson, net- ware components; it plans to make money she asserts. work technology analyst with Wessels, off the software code it puts in the switch. Arnold 8: Henderson, states, “The com- According to Gartner’s LeBaron, Ipsilon’s Competition bined router, LAN switch, and hub market strategy is very aggressive and ambitious. But speaking of pressure, M ETA’s Sribar in ’95 was about $10 billion. The IP switch- “It’s trying to be the Microsoft of routing,” states, “Ipsilon has to stay ahead of Cisco ing business,like LAN switching,could be- she claims. “To do this, Ipsilon is seeding and the other big guys.” Tom Lyon admits come a multibillion-dollar market over the the market with as many software ports as this, but says of Cisco, “They own most of next three-to-five years if the concept it can. The goal is to populate the world the IP market and are most threatened by catches on.” with IP switching software on everybody’s the price/performance advantage of our Market hardware, and collect royalties.” To foster switch.” Ipsilon’s switching solution poten- this population explosion and expand the tially makes a lot of Cisco’s hardware go “The main challenge for Ipsilon right now,” annuity stream, Ipsilon is partnering with away. So Cisco has to gure out a way to im- says META’s Sribar,“is to get the product those mainstream companies which are plement a competing solution, while re- out the door and create a clear identity for especially open to implementing its taining the iron in its installed base. “It’s a the IP switching market.” To speed up the switch. bit of a mindshare play right now,” says process, Ipsilon uses off-the-shelf hard- The strategy involves Ipsilon licensing its LeBaron. “Cisco has red the rst shot AUGUST 1996 THE H_§RR1__Ne
3. COMPANY PROFILE across the bow with a modied router it claims does the same things as the Ipsilon switch. This is designed to make Cisco’s customers feel secure with the networks W 1 Q they already have.” Could Cisco or one ofthe others buy the =1 IP switching solution by buying Ipsilon? Tom Lyon adamantly denies this,“We don’t 5 ”§ t the market message, product, or busi- ness model of any of the Big Three. We 1%; don’t see ourselves as buyout bait!” LeBaron seems to agree, at least concern- % g ‘ ing Cisco. “Cisco could re-create the soft- ware far less expensively than it could buy Fr? tit‘ t Ipsilon. But for the short term, it doesn’t t Cisco’s business model to do either.”Mean- while Cisco does appear to be keeping a prove that its IP switch will scale to handle close eye on Ipsilon. According to Larry heavy network trafc. The grand test is at Blair, Ipsilon’s vice president ofmarketing, hand. Mohr Davidow’s Feiber agrees, “Ip- l2% of the hits on Ipsilon’s Web site come silon must execute by implementing on from Cisco employees. large networks.” LeBaron is“cautiously op- timistic,” although she is condent that Ip- silon can x the technical gotchas along the It might seem promising to throw out all the way.“Ipsilon’s big idea is alreadylarger than unnecessary iron in favor of Ipsilon’s sim- life,” says Feiber. “The challenge now is to ple black box implementation. But will re- live up to all the enthusiasm.” ality match theory? Tom Lyon didn’t shave Written by Michael C. Perkins (michael@ his head for nothing; Ipsilon now has to herringcom). The grand test IPSILON AT-A-GLANCE A A Address: 2191 E. Bayshore Rd., Palo Alto, CA 94303 ' Phone: 415/846-4600 _ Fax: 415/855-1414 ,1 A Heb: www.ipsilon.com till-“irate Established: October 14, 1994 1*Blinker of Employees: 60 i I 1 Technology: IP Switch—designed to combine the intelligence and 1 control ofIP routing with the high speed and capacity 1 , ofATM switching into a single scalable platform. 1 » Product: Ipsilon IP Switch ATM 1600 ;*l.~&n;p rmluut ieteased: April 1996 if , g * A scal Year End: Ianuary 31 Last Financing: April 1996 fiiilh}, of ‘Ljastliiinanc-ing: $22 million 7i-Lapel, Tapital Raised: $32 million J1 ' . 1 A Investors: Merrill, Pickard, Anderson 8: Eyre, Mohr, Davidow A A Ventures, Sutter HillVentures, Digital Equipment Corp., Greylock Management, Amerindo ,lntegrated Device Technology, Integral Capital Partners, First Analysis , ' lg Corporation ‘ Expected 1P0 hate: Calendar 1998 I__H,,,E,ll.R..§.P_ H4» E R.F<_,_l__N_§ A u G u s l 19> 9 6