JBoss/Professional Open Source
Flashback: Investing in 'Professional Open Source' - Exclusive 2004 Interview with David Skok, Matrix Partners
Skok Tells Why Investing $10M in JBoss Was Bold But Logical
By: Jeremy Geelan
Apr. 15, 2004 12:00 AM
This past February David Skok's new VC firm - Matrix Partners - orchestrated, with Accel, a $10 million investment in JBoss, Inc. This first round of funding in an open source company was a bold play, but then David Skok, famous in the Java arena as the founder of SilverStream Software - acquired by Novell in 2002 - is no stranger to bold moves.
The first question then, naturally enough, is to ask Skok whether this investment is somehow a contradiction, or whether there is indeed money to be made from open source?
"We think that there's definitely money to be made," Skok says without hesitation. "Red Hat's a clear example," he adds. "They're now profitable on around a $100m annual revenue run-rate. And we're still in the early days of the open source movement."
Curious about the business model he sees underlying OSS, I ask him how he usually explains it to fellow members of the investment community.
There are two business models, Skok explains, "dual license" and "support."
"In the dual license model," he continues, "followed by companies such as MySQL, software is available free with a restrictive GPL license, and available for a small fee with a less restrictive commercial license.
"You also see companies like SourceFire," Skok adds - they are the ones who developed Snort, the network intrusion detection system - "offering an upgrade product to Snort for a fee."
Then there is the second model, the support model. "JBoss follows the support model," Skok says. "What they call 'Professional Open Source'."
Customers Want a 'Throat to Choke'
Is "Professional Open Source" the wave of the future? I ask.
"Open source should be thought of as a different way of developing software," Skok replies. "With Professional Open Source, we should stop seeing open source developers as unemployed student types wearing Birkenstocks. Specifically, open source for middleware software makes a lot of sense because it's the most open/naked state of code and the peer review process leads to very robust, fast, and open stacks."
Given that open source is fast becoming a multimillion-dollar - most likely a multibillion-dollar - business, I ask Skok who the stakeholders are who can expect to benefit.
"I think that the greatest beneficiaries are customers and independent software vendors (ISVs) who will benefit from high-quality, royalty-free platforms," he replies.
Skok then adds a thought about profitability, for open source and profit are not mutually exclusive it seems. "One of the things that impressed us most about JBoss," he says, "is that it has been profitable since the beginning, and expects to continue operating that way into the future."
The $10 million capital was primarily raised to strengthen their balance sheet, Skok explains, reassuring customers that they are dealing with a company that is here for the long term. "They will also use the capital to expand operations into more regions and scale up support capabilities," he adds.
On a technical note, while JBoss has always been applauded for being robust and scalable, it's never had a GUI-based IDE, thus leaving the mass marketplace solely in the hands of closed source competitors. So I ask Skok what he thinks about the recent idea of plugging JBoss into the Eclipse development framework, making the product a legitimate option for the thousands of developers who prefer a GUI. Will that transform JBoss's fortunes from niche player to mass-market player, does he think? Does that partly explain the timing of this investment?
"In today's market," Skok answers, "it's no longer necessary to get GUI tools and an app server from one company. In addition to the Eclipse project, JBoss is already integrated in all leading IDEs: JBuilder X, Compuware OptimalJ 3.0, Eclipse, IntelliJ, and so on.
"I keep speaking to new startups that are coming out with hot new GUI-based development tools for things like UI development, and building XML services-oriented applications," he continues, "and they all tell me that they are planning to ship with JBoss as the default application server. They tell me that they initially picked JBoss because it's royalty free, but they then grew to love it because it's highly stable and has a very compact, modular architecture.
"The surveys that we discovered during our due diligence," Skok adds, "showed that JBoss was number one in terms of both developer and OEM/ISV usage, and had grown its enterprise market share to 27%. In the last year, they had grown their market share more than twice as fast as either BEA or IBM. They have managed to attract an incredible base of ISVs including companies like Apple Computer, Sterling Commerce, and Mercury Interactive to name just a few. Those were some of the key factors behind our decision to invest."
'Disruptive' Open Source Is the Future
"We believe that their business model is the right one for today and the future," he maintains. "It's totally in tune with what customers want, which is low ownership costs for software, and a fully accountable vendor for support."
It's also a very nice business model financially, he adds, as support contracts are generally renewed, "so you end up with a very predictable revenue stream that only grows over time."
However, as with any business, Skok points out, "It's important to keep listening to the customers to see if there are tweaks that are needed going forward."
What about the wider picture, beyond JBoss - does Skok expect a wave of VC money to now pile into open source in the next five years in the same way as it did, say, to the wireless space in the past five years? Or does he think it will be an altogether more cautious process, more on a case-by-case basis?
"Open source is very hot right now," he says, "driven by the clear success of Linux in the enterprise. We have IBM to thank for that; I've heard that they've committed $2 billion to Linux marketing."
Wishing to further explore the whole Linux phenomenon, I ask him why he thinks it has taken off so astonishingly. He explains that the key to it, according to the customers he speaks to, is the financial attraction of escaping lock-in. "Customers tell me that they're attracted to Linux because of 'Intel Economics' - the ability to ride the price curve of Intel-based computing, as opposed to being locked into proprietary Unix platforms like Sun," Skok says.
"The initial open source successes are Linux and Apache," he continues. "However, customers are looking at the rest of the infrastructure stack and realizing that there are the same cost and lock-in avoidance benefits to be had."
Skok ranks JBoss as one of the four greatest open source opportunities right now: "The two leading companies, behind Red Hat and SUSE, that will benefit next are JBoss and MySQL. There will be others, but it's my belief that only a few will be big enough for venture capital style returns."
What about the whole open source versus proprietary debate: Does Skok think it's operationally - and legally - feasible to commingle the two types of software? Or does he fear like many that we're just ushering in an era of energy-sapping litigation, as IP issues are contested by teams of lawyers in the courtrooms of America and indeed the world?
There are a couple of main bumps in the road, Skok reckons: one is the GPL and the other is the whole question of intellectual property. "Users of open source need to watch out for the GPL license, which can require that they open source their own software under certain circumstances," he notes. JBoss, Skok adds, uses LGPL (Lesser GPL), which eliminates this problem.
The SCO lawsuit, says Skok, highlights the second issue - whether you're protected from IP problems. "JBoss solves this problem by indemnifying its customers against those liabilities. Novell and HP have chosen to do the same thing for their customers buying Linux."
What does JBoss, Inc.'s, newfound "Professional Open Source" status do to the whole landscape - do proprietary app server giants, like IBM's WebSphere, BEA's WebLogic, and Oracle's IAS 10g need to worry that JBoss will make them redundant, I wonder.
"JBoss (and open source in general) is highly disruptive to the business models of those vendors," Skok replies. "The recent recession put an extreme focus on costs, and forced IT to look for ways to do more with less. In the late '90s it was cool to tell your friends about the latest Sun E10k box that you had purchased for over a million dollars. Now it's cool to tell them how much money you've saved your company by deploying Linux on commodity Intel boxes."
Skok tells me how he's spoken to several customers who had initially chosen WebLogic or WebSphere as their app server of choice, and then switched to JBoss "because of the extraordinary cost savings," as he puts it.
"Their developers had all recommended JBoss as their preferred product," he adds. "It took a budget squeeze for them to take a chance and try it, but once they saw how good the product was, they weren't going back."
Technology Startups: Then and Now
"When we founded SilverStream, it was right when enterprises were starting to use the Web for business purposes," he recalls. "There was untold excitement and promise, and the future was wide open.
"Enterprises felt a lot of pressure to move quickly or be 'Amazoned.' That led them to ignore their traditional purchasing habits and buy from startups to gain a time-to-market advantage.
"After the bubble," he continues, "the pendulum swung too far the other way. No one wanted to buy from startups, and all the shelfware had eroded trust in sales promises of magical business gains. The pendulum is starting to swing back to normal, but it's still an extraordinarily tough environment. Products have to provide really clear demonstrable value and significant ROI in a short period of time. Lots of reference-able customers are crucial, and frequently startups need a channel partner to help solve the 'safe choice' problem."
In those days, Skok recounts, it was essential to build market share quickly, at any cost: all focus was on revenue growth, and profitability was considered very unfashionable. "These days," he says, "the focus is on capital efficiency, building businesses with the minimum amount of capital." That means a strong focus on expense control. Rushing to get to market is less important than really getting the product right and making the customer totally satisfied.
"We're back to the days where it will take four to five years to build a real company, and the only people starting them are those with a true passion for the technology and their customers," adds Skok. "I've always believed that the only way to make money is not to focus on making money, but to focus on building real value and making your customers happy. In the bubble, we had lots of the wrong kinds of people starting companies, trying to make a quick buck."
So much for what's different. But there must be similarities too, "eternal truths" of the startup game, if you will.
Skok agrees. "What hasn't changed is that you still need to focus on building an exceptional management team consisting of only A players," he says.
"A players attract other A players; B players attract C players. The difference in performance between an A player and a B player is enormous, whether it be development, sales, or any other function.
"Over many years," Skok says, "VC firms have collected a lot of wisdom about what it takes to succeed. The key ingredients that we look for are:
"As a parent you have constant day-to-day involvement with the child including all the highs and lows (changing nappies, etc.). As a grandparent, you're less involved, get to share in some of the fun, but also have less of the day-to day-stress. The biggest change is switching from making all the decisions to becoming an advisor and mentor to the CEO."
So he doesn't object when I ask him to name one thing that Java has gotten completely right - and, conversely, one thing in Java's history he wished had turned out differently.
"Java got two things right," Skok says. "First, platform independence - which means the same app will run on any platform unchanged. And second, standardization - which means vendor choice and competition."
What Java has gotten wrong, in his opinion, is that "it has become too complex."
"That complexity is necessary in certain respects (as distributed applications are complex in nature). Expert developers appreciate the need for this. However, not enough attention has been paid to simplifying development, and making Java more accessible to a less expert audience who I believe make up at least 60% of the corporate developer audience."
The most promising answer to the over-complication of J2EE, Skok adds, is something JBoss introduced in their 4.0 release: aspect oriented programming.
"AOP has the potential to greatly reduce the complexity, allowing developers to add capabilities like persistence, remoting, security, transactional integrity, and so on, to plain old Java objects (POJOs) without the complexity of today's EJBs. This is close to the .NET approach. JBoss is working hard on the EJB 3.0 specification to help bring in these kinds of advances. They're hopeful that that group will find a way to make enterprise Java easier to use. This is important to keep the overall market share of Java versus .NET."
In addition, Skok believes that the J2EE world needs high productivity GUI tools for less expert developers in the area of UI development. Things to help them graphically design page flows, and build data-bound forms and reports.
"Today .NET is the only serious challenger to Java in the enterprise," he says. "I'm less hopeful for Java on the client so long as Microsoft continues to fight it. But I haven't written it off. I think that we'll continue to see both .NET and Java as competing platforms in the foreseeable future. Web services will greatly increase interoperability."
What's his response to those who argue that Sun has somehow lost its moral right to exercise exclusive control over Java and that Java ought now to be open sourced? "I'm in favor of open sourcing the source code for the Java reference implementations, but allowing Sun to retain the rights to the Java brand. Sun could still maintain control over the verification and certification of products based on them passing Sun's compliance tests. That model will dramatically improve the quality of Java, give developers the comfort that they are not playing in a totally Sun-controlled world, and have the ability to contribute to the reference implementation, while still leaving Sun a way to make money out of Java.
"In my opinion that would increase the acceptance of Java in the market, and reduce the chances of success for competitive languages like C#."
Beyond Java: Technologies to Watch
"I'm personally very excited about the opportunities around the datacenter," Skok explains. "It's clear that today's datacenters are way too complex to manage, and can't respond fast enough to changing business needs."
The datacenter of the future, he believes, will be very different: there will be a big pool of storage, a large pool of processors, and a big network pipe coming into a lights-out facility.
"You'll only need to go in there to add new processors/disks or to remove broken units. New virtual computers will be created out of these pools via a simple GUI. These virtual computers will share resources from the pools, but will be isolated from one another. They will be dynamically allocated more or fewer resources as demand grows and shrinks. My investment play in this area is a company called Katana."
"RFID is also very hot right now," Skok says, "driven by a mandate from Wal-Mart that its top 100 suppliers have to start shipping them pallets that are identified by RFID tags.
"This was followed by a similar mandate from the Department of Defense. That has kicked off a mad scramble by their vendors to become RFID enabled. We've invested in a company called OAT Systems, and they are overwhelmed with customer demand."
Infosecurity is another interesting area, he says. "One of the driving forces is regulations such as HIPAA, GLBA, etc. We're seeing a fair number of business plans focused on auditing and protecting access to information assets in databases, files, and so on."
Skok has also spent a lot of time investigating grid computing. "I'm less enthusiastic in the short term, though," he says. "Grid computing today works best for technical computing applications where the workload can be easily split into modules that can be run remotely in parallel. It also requires that applications be rewritten to support a Grid API (OGSA being the key standard).
"When I look at enterprises, I see mostly applications that cannot be run remotely as they all require access to a single database, and I see a huge reluctance to rewrite applications. What's more interesting to enterprises is the utility computing model where they can run these applications unchanged on a shared infrastructure that responds dynamically to changing workloads. That will be a much bigger market in the short to medium term."
Over time, grid computing could become more interesting, he conceded - "as we move to highly modularized applications built out of Web service components."
Skok ends by stating how VCs feel about the current environment, after going through the tech recession of the last few years.
"There's a definitely increased optimism right now. We're starting to see companies meeting or beating their plans once again. As an example, one of my portfolio companies, Netezza, had a very aggressive forecast for its fourth quarter, and was able to bring in bookings that were double that target. We haven't seen that in a long time.
"The public markets are getting better, and will be helped by the IPOs of Google and Salesforce.com. And there is a lot of M&A activity.
"Off-setting that positive news," Skok adds, "we still have an environment where there is too much venture capital chasing too few deals; an overhang from the bubble. This means that there is overfunding in hot areas. As an example, there were no less than 64 companies building WiFi chips when we last counted. I predict that in the long run only the top-tier venture firms will make money and survive. They have access to the best deals, and have the most experience in helping those companies build for success and avoid pitfalls."
Finally, is there such a thing as an "ultimate software solution" - or does Skok view that as a mere mirage? "I guess it would have to be a system where you could think at it, and have it interpret your thoughts to build fully functional, scalable, reliable, secure applications automatically," he says. "It would also want to behave more like biological systems: i.e., be autonomic (self-healing), able to detect changing needs in the environment, and suggest ways to evolve itself.
"Is that a mirage? For the time being, I guess so..."
Reader Feedback: Page 1 of 1
Latest Cloud Developer Stories
Subscribe to the World's Most Powerful Newsletters
Subscribe to Our Rss Feeds & Get Your SYS-CON News Live!
SYS-CON Featured Whitepapers
Most Read This Week
Breaking Cloud Computing News