Wednesday, February 09, 2005
Netezza gets $15-million Series D for streamlining queries
How to deal with a mountain of data: divide and conquer.
February 7, 2005 Print Issue
Netezza, which sells integrated processing, memory, and storage devices designed to improve database queries, recently closed a $15-million round of funding, the company’s fourth. Meritech Capital partners led the round, to raise Netezza’s total funding to $68 million since its inception in 2000. According to CFO Patrick Scannell, the money will go toward increasing sales, expanding support services, and moving into the federal market.
Previous investors including Battery Ventures, Charles River Ventures, Matrix Partners, Orange Ventures, and Sequoia Capital returned for the round.
Why it matters
The global market for storage hardware and storage software exceeded $20 billion in 2004, according to Stephanie Balaouras, an analyst at research firm the Yankee Group. But data, in itself, has little value. Being able to resolve hundreds of millions of points into one big picture suddenly morphs the billions spent on storage disks, servers, and databases into a competitive advantage.
Knowing what the database knows has never been easy. Traditionally, the storage server would scan all of its data like a person reading a book, starting at the front and moving straight through. As growth in available storage space exploded, processors could not keep up. Netezza breaks down the single processor bottleneck by breaking the database down into manageable parts and giving each part its own processor. The company’s smallest product spreads 400 GB of data across 60 processors. The result: faster database searches.
“Things that would take hours, or even days, of crunching take minutes,” says Arun Taneja, the founder of analyst firm the Taneja Group. “There’s a gazillion business intelligence applications for this.”
CEO Jit Saxena founded two companies and successfully secured exits for his investors. He took Applix, his first company, public in 1994 for $18.5 million—eight years after his last infusion of venture capital. The company boasted a nice initial pop in price, jumping 50 percent in less than two months. The company’s market capitalization sat at over half a billion dollars in 1997; but since then, Applix has fallen apart. In 2003, the company lost $10,000 on $27 million in revenues and its stock now sits at around $4.
“This is a very old problem that a lot of companies have tried to solve,” says Mr. Taneja. Fortunately for Netezza, nobody has quite solved it yet. However, several large companies have good reason to keep the startup from ever gaining traction. There’s IBM, the undisputed king of data; Teradata, the go-to arm of customer relationship management company NCR; oh, and then there’s Oracle, whose mad-dog CEO Larry Ellison is apt to squash a company for even looking like it wants to compete. “That wouldn’t be a good thing for any startup,” says Mr. Taneja.
Despite a daunting list of competitors, companies that make business information easier to pull from customer data have done well in the past. Seisint, for example, had a technologically similar product. The company partnered with its customers, signing Equifax, Bank Atlantic, and LexisNexis as investors. LexisNexis liked what it saw enough to buy the company for $775 million.
Netezza CFO Patrick Scannell says the company now has 17 customers and has sold 40 systems at an average price of $1.5 million per system since it started shipping products two years ago. So far, the company has not secured any of its customers as investors. More importantly, Netezza is not profitable and doesn’t anticipate reaching profitability for at least another six months.
The company touts its low cost of production, but if it can’t turn a profit, something’s not working right. Still, if Netazza can deliver the 50x acceleration in database queries it advertises, it may quickly find itself profitable. Profitability will attract the attention of the big database companies and Netezza may want to look for strategies to turn its competitors into potential partners, or even potential acquirers.
Mazu Networks raises $12-million Series C for intrusion-prevention software
Mazu polices network traffic and can pull over digital attacks before they get out of hand, but will it go unnoticed in the security market?
January 24, 2005 Print Issue
Mazu Networks, based in
Why it matters
Mazu, founded in 2000, strives for the holy grail of security: protecting networks from completely unknown attacks. The company’s software monitors data flow, looks for specific patterns, and sends an alert when it recognizes abnormal network traffic.
Mazu’s Enforcer network monitor sits in an interesting piece of market real estate. The company calls its monitor an intrusion-prevention engine, but the product only watches the network for changes in data transfer rates and sends alerts. Mazu markets its technology with the promise of preventing new types of threats before they emerge by watching for traffic that looks similar to previous attacks.
A recent study by the Computer Security Institute showed 45 percent of IT professionals use some sort of intrusion-prevention system and 68 percent use intrusion detection. Mazu’s product does not prevent attacks, but it does detect them in a novel way.
Symantec voted for Mazu’s technology with its pocket book. The security and now storage software giant has been known to make strategic investments in companies it wants to acquire.
Max Poletto and Eddie Kohler formed the company as postdoctoral researchers at MIT in 2000. Since then, the company has expanded to 50 employees and Don Casey has taken the helm as CEO. Mr. Casey formerly ran networking divisions at IBM and Apple.
Mazu plays in a crowded market. Arbor Networks, based in
Both companies compete against a slew of other security startups and established vendors trying to cash in on the $2-billion enterprise network security market. Mazu has a compelling remedy for certain attacks, but by no means provides a silver-bullet solution.
Tom Corn, vice president of business development, says the new funds will help expand the sales channel, something the company will need to compete against the plethora of security solutions stemming from both the money of overly eager venture capitalists and the security anxiety of large IT companies. Signing Symantec would have been a big step toward boosting sales quickly, but with the recently announced buyout of Veritas, the security giant may not be able to offer attention to startups.
Successful security exits aren’t that uncommon these days—as 3Com’s recent $430-million acquisition of intrusion-prevention device vendor TippingPoint testifies—but this deal has a limited upside. Mazu doesn’t have the easily understood business model or compelling market to score an excellent IPO. No matter how hungry the company is, the slice of the pie it’s fighting for won’t be big enough to sate investors’ appetites.
VC Action: Chelsio Communications gets $25-million Series C for ethernet acceleration
Exploiting ASICs to boost networking speeds may pay off—but not this year.
January 31, 2005 Print Issue
Chelsio Communications secured a $25-million Series C to continue development and expand sales of its ethernet acceleration cards. INVESCO Private Capital led the round with Sequoia Capital, New Enterprise Associates, Global Catalyst Partners, Pacesetter Capital Group, Horizon Ventures, and Abacus Capital also anteing in. The latest round brings Chelsio’s total to $55 million.
Why it matters
If Chelsio succeeds, it will be for exploiting application specific integrated circuits (ASICs) to boost networking speeds. While the chips used as CPUs juggle many tasks, ASICs do one thing and do it well. This specialization can have huge benefits.
For example, dedicated graphics processing chips have taken the eye candy in gaming consoles and PC games to entirely new levels over the past decade. Chelsio plans to do much the same, by out-grinding more generalized chips to complete networking tasks much faster.
As web-based enterprise applications and storage devices go online, quickly processing network communications becomes critical. Before a computer can send information over the network, it must first translate the data into networking language. TCP/IP sets rules by which an application makes data transferable.
Generally, TCP/IP requires effort from the CPU to translate network communications. Chelsio sells a specialized processor exclusively for executing the TCP/IP protocols, which takes strain from the CPU and allows applications to run faster.
The first attempt at making TCP/IP-exclusive processors failed miserably. Companies designed the processors to operate on 1-GB -per-second networks, but improvements in CPU speeds rendered their ASICs obsolete. As a general rule, it takes one processor cycle, or hertz, to translate communications of 1 bit per second.
As Intel and AMD drove CPU speeds into the 2- and 3-GHz range, computers had plenty of resources for running applications and translating communications from the 1-GB-per-second network. Companies such as Alacritech, Qlogic, and Adaptec took a big hit.
Chelsio may be able to avoid the pain of past ASIC attempts. It has designed its product for 10-GB ethernet, the next quantum leap in network speed. Since AMD and Intel have failed to deliver on
The company, founded in 2001, now has 50 employees. Much of the management team has a Silicon Graphics heritage. CEO Kianoosh Naghshineh put in time as CEO of ASIC Designers, which designed and marketed the intellectual property behind these specialized circuits.
Chelsio will be competing with a slew of startups hoping to capitalize on a migration to 10-GB ethernet, and a couple of walking wounded from the 1-GB fallout. Players include private companies such as Alacritech, which has raised $35 million since its inception in 1997; Silverback Systems, founded in 2000, with $32.3 million in venture funding; Siliquent, which has scored $31 million, starting in 2001; and S2io, which has pulled down $42 million worth of funding in the last three years. Publicly held players include Adaptec, with more than $470 million in sales last year; Qlogic, which booked almost $530 million; and Intel, which sold $33.35 billion last year.
Each player offers slightly different functionality and features on its ASICs, but the competition will most likely come down to cost. The first to market with low-cost silicon may get the pop needed to establish long-term leadership.
Success hinges on 10-GB ethernet adoption in the data center. If storage networks and enterprise application servers migrate to faster network architecture, the computers will need processing muscle to keep up.
Mr. Naghshineh says he expects the company to become profitable in early 2006, but to get the product adopted, Chelsio will have to simultaneously boost its sales expenditures and cut its production costs. Reducing margins will be tough, especially when 10-GB hasn’t taken off yet.
If it takes until 2006 for 10-GB ethernet to move from the research lab to the enterprise, Chelsio will need every penny of its $25 million just to stay afloat.
VC Action: AirTight Networks gets $10.25-million Series A for wi-fi security
Data-laden air needs protection, too.
November 29, 2004 Print Issue
AirTight Networks inked a $10.25-million Series A funding round led by Walden International, with additional funding from Trident Capital, Granite Global Ventures, and Blueprint Ventures. CEO David King says the money will be used to ramp up sales in the
Formerly called Wibhu, the company was founded in Pune, India in 2002. It started out developing software to map the coverage area of wireless local area networks (WLANs), but has since focused on wi-fi security, offering intrusion prevention software and monitoring hardware.
Why it matters
By 2008, wireless access networks will be a mainstay for half of U.S. enterprises, according to the Radicati Group. “Corporate air contains data,” says Jai Rawat, vice president of business development for AirTight. “The air is now an asset you must monitor.”
The market for wi-fi security products is flourishing, according to Wai Sing Lee, an analyst at Frost and Sullivan. Mr. Lee expects the market size to double from $100 million in 2004 to $200 million in 2005.
The first wireless security companies offered limited intrusion detection software which plagued systems administrators with false-positive alarms. AirTight claims its software will automatically shut out intruders from the network, search and destroy rogue access points, and pinpoint the location of malicious machines.
AirTight developed 90 percent of its technology in India under the direction of co-founders Pravin Bhagwat and Kiran Deshpande. Mr. Deshpande formerly managed 3,500 employees as CEO of Mahindra British Telecom, and CEO Mr. King formerly ran Proxim, a publicly-held WLAN manufacturer based in Sunnyvale, California.
AirTight’s main competitors got a jump on the company, and have been selling security products for the last two years. Of the companies that reported their revenues to Frost and Sullivan in 2003, Sunnyvale’s AirMagnet had 46 percent of the market, AirDefense in Alpharetta, Georgia claimed 24 percent, and Network Chemistry, in Palo Alto, California, had about 5 percent.
Competition for enterprise wi-fi security will only intensify in the future, and AirTight is a new player in a market of established players. Network Chemistry says it has been profitable for the last three quarters, AirDefense claims five quarters of profitability, and AirMagnet cited eight. AirTight has yet to even ship its product.
“If you define the market as only intrusion detection and marketing, sure, we’re late,” admits Mr. King. But he believes the market should also include intrusion prevention and location mapping, an area where he believes AirTight has a clear advantage.
Anil Khatod, CEO of AirDefense, says the wi-fi security market “will be irresistible for the large guys.” If big acquirers show interest, great technology will take center stage and attract the best valuations. Cisco, for example, doesn’t have any problem logging sales. Market share momentum may not be as important as sustained, low-cost development and functional improvements in this rapidly evolving space.
“Are we late to the party? Absolutely not,” says Mr. King. “The real party’s just started.”
Tuesday, February 08, 2005
VC Action: IronPort gets $45-million Series D for email appliance
Is there life beyond spam control and anti-virus?
November 15, 2004 Print Issue
IronPort Systems, based in San Bruno, California, raised a $45-million Series D round of funding led by New Enterprise Associates, for a total of $90 million – a hefty sum for an appliance company that started its sales just two-and-a-half years ago. CEO Scott Weiss says he will spend about 40 percent of the money ramping up sales, and set aside the rest for future projects.
Why it matters
IronPort, founded in 2000, offers a network-perimeter appliance that uses software from Brightmail and Sophos to identify spam and viruses before they infect the enterprise network.
According to market research firm Gartner, 80 percent of North American and European enterprises will purchase a spam filter by the second half of 2005, and they’ll spend just under $200 million on filters this year. By 2008, sales of appliance-type solutions, such as IronPort’s C60 product, will dominate the market by 10 to 1, edging out other spam solutions, including the use of software inside the network and outsourcing email to a third party.
IronPort is betting on a “razor and blades” strategy: if it can capture market share for the anti-spam and anti-virus capabilities of its email-processing platform, its profit will come from expanding functionality and adding new applications.
After helping to found Hotmail, IronPort CEO Scott Weiss headed to IdeaLab, where he ran into Scott Banister, who had worked for an email list-hosting company. They started IronPort to develop a better email-processing appliance. Today, the company employs 210 people, and just added former NetScreen CEO Robert Thomas to its board.
In the appliance market for email management, it’s a battle between IronPort and CipherTrust. CipherTrust, based in Atlanta, was founded in 2000 and just secured $42 million in its first round of funding in March.
Market research firm IDC reported in August that CipherTrust garners up to 20 percent of the content-management appliance market – three times more than IronPort. IDC said that in 2003, CipherTrust had $26 million in revenue compared to IronPort’s $8 million. Mr. Weiss is skeptical of these numbers, saying IronPort has “ripped out CipherTrust boxes for at least three big customers.”
If email management expands beyond spam and virus protection, the company that already has a box on the network could end up sitting pretty.
But IronPort’s razor and blades strategy, with its multiple partners, is a big risk – one that its competitors aren’t making. CipherTrust has done its email-feature development in-house, which CEO Steve Raber says saves money. If each partner holds out their hands for a piece of the sale, suddenly there isn’t much left for IronPort.
Although Peter Morris, a venture capitalist with New Enterprise Associates in Menlo Park, California, says IronPort is “just hitting the cover off the ball,” the company’s revenue growth has been tempered by its licensing commitments.
Mr. Weiss insists he isn’t being held hostage by his partnerships. He prefers to keep development costs low, focus on the next level of email management, and develop new services that the competitors don’t have.
IronPort should be profitable by mid-2005, says Mr. Weiss, but adds that he “may push out profitability to get market share.” He may be able to forestall profitability in favor of bookings expansion for a while, but IronPort has to make a profit eventually.
As CipherTrust’s Mr. Raber says, “When you run out of VC money, you’ll really believe profitability is important.”
VC Action: Optinuity gets $6.5-million Series A for automated IT management
Why buy another fire alarm when what you really need is a sprinkler system?
December 6, 2004 Print Issue
Optinuity, based in
Optinuity makes software that CEO Scott Stouffer claims will automatically do the work of IT experts. If he’s right, then companies with Optinuity software will no longer have to page their IT staff when something goes wrong with computers. Instead, Optinuity will bring the systems back up and running and run the necessary maintenance procedures without the need for human intervention.
Why it matters
IT staff members spend between 50 and 80 percent of their time “performing maintenance and repetitive, predictable tasks,” according to Richard Ptak of analyst firm Ptak, Noel & Associates in Amherst, New Hampshire. Automating what would otherwise require human interaction with a faulty system could result in fewer human errors and machine-speed execution – this is an attractive proposition for companies such as banks that lose millions of dollars per minute of computer downtime.
Just as the assembly line reduced the demand for manufacturing jobs, Optinuity’s software, if it works, promises to reduce the enterprise need for IT staff. When a network goes down, one staff member can press a button and an automated series of protocols will execute to bring the system back online. Enterprises will no longer have to rely on experts to cobble together specialized mini-programs to help them retrieve functionality quickly or pay them above-market prices for their proprietary, onsite training. Perhaps most importantly, “This will make enterprises less dependant on waking up that guy with the pager,” says New Enterprise Associates venture partner Rob McGovern, an Optinuity investor.
Before he was Optinuity CEO, Mr. Stouffer founded Visual Networks, which he grew to a $100-million IT-management software vendor. The technology founders are Rachid Sijelmassi and Anil Parthasarathy, both computer scientists. The company employees five, but Mr. Stouffer says he hopes to hire 15 new employees by year’s end.
This nascent market may quickly attract the attention of the large systems integrators who have an easy entrée into the system through their event alarm products. “Right now, there are lots of companies that make the fire alarms,” says Mr. McGovern. “We believe Optinuity is the first to make the sprinklers.” Software vendor BMC in
Optinuity’s greatest asset may prove to be its management. Mr. Stouffer has a track record of selling network tools that help IT experts make decisions about rerouting congestion and solving information traffic problems. “It’s an interesting challenge to get people to understand what this thing does,” says Mr. Stouffer.
In a room full of 3Com switches, Cisco servers, IBM databases, and Dell terminals, it’s easy to overlook a software startup. “They’ll have to be very clear with how they can help,” says analyst Richard Ptak.
VC Action: LinkedIn connects with $10-million Series B for social networking
It’s all about who you know.
October 13, 2004
LinkedIn, an online social networking company, announced today that it received $10 million in Series B funding. Venture Capitalist firm Greylock led the round, raising LinkedIn’s total funding to $14.7 million. LinkedIn co-founder and vice president of marketing Konstantin Guericke said his company would use the money to expand internationally, improve customer service, and invent new features to add revenue.
LinkedIn, based in Mountain View, California, now boasts 1.2 million registered users, up from 40,000 less than a year ago. The LinkedIn network expands by invitation only, and has been marketed to business people since its inception in May 2003.
Why it matters
Venture capitalists have been throwing money at social networking services for slightly more than a year now. Tribe.net got $6.4 million last November, led by Mayfield. Friendster pulled in $13 million from its Series A last October, led by Kleiner Perkins and Benchmark. US Venture Partners put its weight behind Spoke Software, for a $4.2-million Series A, and the Partech’s $5-million series B. Most recently, Spoke also closed an $11.7-million Series C led by Doll Capital Management.
Each of these services operates on the same basic principle and design. One person signs onto the network, and invites three friends to follow. Each of the friends invites three friends, and so on, until seemingly the whole world is using the service. Each person in the network is linked to others in the network. Alice might not know Catherine, but she is good friends with Bob, who went to high school with Catherine. So Alice can contact Catherine through Bob.
Where Friendster attracts people looking for friends, LinkedIn’s management has targeted its service at business people looking to pump up their Rolodexes. You may not know a Java programmer, but your system administrator might, and if both are signed up for LinkedIn, you may be able to make that connection.
LinkedIn’s CEO Reid Hoffman, who came from an executive vice president position at PayPal, sits on several high-tech boards and has been an angel investor in at least three recent startups, including Ironport and Nanosolar. CTO Eric Ly co-founded two other companies before founding LinkedIn. Jean-Luc Vaillant, LinkedIn’s vice president of engineering, has over 15 years of software design under his belt.
LinkedIn and Friendster serve different markets. Mr. Hoffman lists himself as one of Friendster’s angel investors as if to draw a clear line between networking for work and networking for fun. The closest competitor to LinkedIn in the business space is Spoke, which sells enterprise software designed to help salespersons find the right contacts to call within a company. Instead of adding names through invitations, Spoke connections are added through a company database. Company Vice President Mike Trigg said its product was a “sales prospecting tool” rather than a social networking service. He also said that some Spoke users also used LinkedIn to mine sales opportunities.
Many of the initial bugs in social networking have been worked out; personal contact information can be restricted so that executives can get information without giving away their Rolodexes. The idea is catching on, but the question on everyone’s mind is: can social networking sites make money?
LinkedIn currently scrapes together revenue through ad sales of targeted, search-specific, sponsored links. It has partnered with Market Banker to sell text ads about 10 words long for $595 per week or $295 for three days. Mr. Guericke said he expects the company to reach profitability sometime in the second half of 2005.
David Sze, a venture capitalist at Greylock and a LinkedIn board member, said, “A subset of folks will see a huge value in this and be willing to pay a substantial amount for the service.” Mr. Guericke, one of LinkedIn’s co-founders, said the recent round of funding would be used to tap recruiters - who use LinkedIn to find executives with expertise - for more revenue. He said the company hopes to develop new features for recruiters, HR departments, and other hiring executives. And those features would come at a price. A successful blend of targeted advertising and service sales could lead LinkedIn down the yellow brick road to profit.
But Charlene Li, a social networking expert at analyst firm Forrester, said that her company recently conducted a 10,000 person survey which found that about 8 percent of North American Internet users have joined some sort of social networking site. Of those signed up, only 2 percent would be willing to pay to be introduced to someone outside his or her personal network.
VC Action: Diligent gets $22-million Series B for virtual tape
The market is heating up for enterprise software that transfers data from magnetic tape to storage disks.
October 7, 2004
Diligent Technologies closed a second round of funding worth $22 million on Monday, giving the
A spin-off of storage giant EMC, based in Hopkinton, Massachusetts, Diligent writes software that streamlines the transfer of massive amounts of data from software applications - such as those used to manage human resource and financial departments - to storage disks. Accel Partners and Matrix Partners co-led the B round, with backup from Gemini Israel Funds.
With a research and development arm in
In addition to accelerating realization of its VTF Open product, which is software that makes data backups easier, Mr. Kempel said the second round will go to developing new features for future Diligent products.
Why it matters
A monumental migration in enterprise data storage is underway. Magnetic tape, the old standby for most companies, is cheap, but unreliable and slow. For the first time, however, disk storage prices have reached a point - after falling 40 percent every year - where using the technology in the enterprise is an option.
What’s the catch? Shifting from tape to disk is not easy. Backup policy must be rewritten, and reprogramming applications to recognize a new storage format is costly and time-consuming. Enter Diligent and its competitors. So-called virtual tape software allows system administrators to gain the benefits of disk storage and recovery without having to reconfigure the data systems.
More importantly, virtual tape will provide an automated architecture for data archiving. Michael Peterson, a storage expert at analyst firm Strategic Research Group, said, “There won’t be one in every sever room, but there’ll be at least one in every data center, and there are thousands of those. People are paying beaucoup bucks for this.”
Each of the company’s executives has experience working for storage companies, but only one, CTO Neville Yates, has ever founded a company before. Yates co-founded virtual tape company Sutmyn Storage, and headed that company’s spin-off, V-Stor. In 2002, ADIC bought V-Stor and leveraged the technology to develop its Pathlight VX virtual tape product. He joined Diligent in July 2003.
The market breaks down into two segments: open system virtual tape and mainframe virtual tape.
The large mainframe suppliers provide virtual tape products such as IBM’s TotalStorage Virtual Tape Server and Fujitsu Siemens Computer’s CentricStor. These products are designed to compliment a specific type of mainframe and have little traction for companies which use other data storage systems.
For companies that run their data through distributed computing or open systems, a wide range of startups have sprouted up to reap virtual tape dollars. ADIC, FalconStor, Overland Storage, Quantum, and Sepaton all sell variations on the virtual tape theme.
Diligent spans both spaces and claims 80 customers, lagging far behind Quantum, which Gartner said can claim the largest open-system customer base, with 150.
Arun Taneja, a storage expert with analyst firm Taneja Group, said, “For the next 18 to 24 months, virtual tape libraries will be the preferred solution.” Strategic Research Group’s Mr. Peterson said, “The next five to 10 years will be a period with lots of chaos in the storage market. Tape will survive and virtual tape will survive to manage it.”
As application vendors write new, disk-storage-based data backup algorithms, virtual tape will move beyond a stopgap solution that bridges legacy systems with newer, cheaper storage methods. It will become the central plumbing that makes data management possible, and cut the time and difficulty of operational recovery.
The more forward-looking startups already are integrating new features that delineate different kinds of data to allow for a variety of potential applications. The next wave of storage software will condition and compress data more efficiently and allow for quicker recovery.
This technology will be around for the long haul, but many of the companies that provide similar solutions will not. New features that facilitate copy and recovery will be important, but with a market size limited to serious data storage sites customer traction will be the key to success. Because Diligent operates as a software supplier to several different platforms, it may be able to leverage its compatibility into securing more customers.