When JBoss pioneered the Professional Open Source Model in September 2003, there were many questions asked:
- Would this make large customers feel safe enough to use open source software for large, mission critical projects?
- Is this a viable business model?
- If this is good, then why is only one company doing business like this?
- Can JBoss continue to gain market share?
- Is JBoss going to be able to continue to expand the community around the project?
- Can JBoss replicate their success beyond the App Server?
Nearly two years later at JavaOne 2005, the record culminated in a series of announcements by JBoss and others that proves the momentum of this model for the Middleware segment of the market. I don’t blog often, but there are enough events for me to try to demarcate this point in the evolution of the Middleware Industry and Professional Open Source. This is really the first time I have written publicly about how our business is going since my first blog describing the JBoss business model in February, 2004. It also seems to be kind of a good time to reflect on the state of JBoss.
Would this make large customers feel safe enough to use open source software for large, mission critical projects?
The answer is clearly yes. We have seen surveys from analyst firms, as well as an increasing number of articles and letters from end users that prove that the number one thing that holds customers back from using open source is Support.
Red Hat and Novell are proving this in the Linux arena. And JBoss success in Java Middleware has been very, very strong in a short period of time.
I am currently in Europe for two months. One of the things I have been doing is meeting with large telco companies – both manufacturers as well as operators. Most of them are JBoss users – rolling out very large scale Network Management platforms controlling thousands of devices, or service delivery platforms supporting millions of users. JBoss is designed in many equipment manufacturers like Siemens, Alcatel, Nortel, Telcordia, etc. These are mission critical systems that feel safe running Professional Open Source projects.
We are seeing similar pick up in government accounts. I recently saw a demonstration of a system built by the Navy to control all of the inventory and logistics for Air Craft Carriers – bringing together all of the resources to send a plane or group of planes on a mission and bring them home safely.
And of course commercial end users are deploying huge applications – Travelocity, Ameritrade, European Bank. Autotrader.com talked at the last JBoss World conference in Atlanta about how they manage 200 JBoss Servers. Norisbank is one of our new European Customer Advisory Board members - they are running €3 Billion per year of business thru JBoss. And there are many that are not willing to go public (although some of the stories are leaking out – like the CIO of Goldman Sachs talking about their growing usage of JBoss, or the GE manager who was bragging about how much money they had saved with JBoss - both at investor conferences). And just as an example of the mass adoption, yesterday an email came in from Pakistan where a person was asking for pricing on a Support Subscription for a banking app that was deployed on 70 servers.
So in terms of both user surveys and real case studies – the answer to the first question is an unqualified YES.
Is this a viable business model?
Fortunately for the customers, partners and employees of JBoss, the answer is yes. JBoss, Inc. is a growing and thriving business. Since the announcement in September, 2003 our monthly sales are up by a factor of about 9X. Our employee base has grown to over 130 people. And we are running about cash flow break even.
In addition, there is a thriving ecosystem that has grown to take advantage of JBoss technology. In a simple calculation of all the hardware running JBoss, all of the software that ships with JBoss (like Mercury or BMC, or Sterling Commerce or Filenet or Adobe or literally thousands of software packages), and all of the System Integrator projects (like the French DGI with Atos Origin, or the State of California with Accenture) – the total is well of $5 Billion (with a “B”). So the whole ecosystem is able to make more money with JBoss.
The keys to the business model are:
- Get mass adoption of our products by having innovative, quality products that are distributed as completely free and open source. As discussed elsewhere we have a strong preference for the LGPL license that assures the software will always be free and open, that others can benefit from changes, and that is easy to bundle in ISV and OEM products.
- Provide Support Subscriptions that have enough value targeted at production users that a growing percentage of that mass user base will do business with us. We get about 70% of our business from this recurring revenue stream. This is the focus of our investments in the JBoss Network that will roll out this year to our customers.
- Partner Ecosystem and Certification for both products that run with JBoss as well as System Integrators who provide the implementation services for JBoss projects. This assures our mutual end customers that there is a tight integration between the products and the companies.
If this is good, then why is only one company doing business like this?
Well, this was actually one of the two sparks that got me to write this up (the other being our JBoss Portal announcement). We are attracting a growing list of companies who are starting to at least try our business model out.
Computer Associates was the first large organization to try this model with Ingres. They went to a total open source model, and focus exclusively on support subscriptions. Since doing this almost a year ago, they have actually seen their business improve. Their customers feel safer now that the code is in open source, and the usage is expanding because of the move to Professional Open Source.
Recently, IBM bought Gluecode to capture the Geronimo project. They state clearly that they will be charging a support subscription and making the Gluecode software open source – as well as trying to upgrade users to the licensed WebSphere product. Sun also announced that they will open source their base reference implementation of their app server – although not yet willing to go the whole way with things like clustering and security. They will focus on selling support subscriptions to this base – as well as trying to upgrade users to a licensed version (which they also package in subscription form now). Sun also announced that the RI for the new Java Business Integration spec will be open sourced, and I am guessing they will be positioning the new SeeBeyond acquisition as an upgraded license sale. Iona also announced Celtix – an open source ESB implementation that they intend to sell support subscriptions for as well as try to sell the licensed version of their product on non-Java platforms like the Mainframe.
In addition, a number of smaller firms are moving to this model. Alfresco announced a nice high end content management system based on an open source model. As well, Panscopic acquired the leading open source reporting project called Japser – and actually renamed their company JasperSoft.
It is interesting to note that many of these offerings involve a dual license strategy. Typically with a low end offering that is open source and a high end offering that is a traditional enterprise model. It is difficult for these vendors to make a shift from the license revenue stream and associated cost structures.
This is all very good for JBoss in several ways. First, it confirms the basic business model of Professional Open Source. Companies standing behind an open source project and being able to fund it by providing support subscriptions on the product. Second, and more importantly, it helps paint the picture for end user organizations as to how to adopt open source. One of the keys to the success of JBoss as a business is to have a larger percentage of our users become customers. If the standard in IT gets set to have support for any significant open source projects, then JBoss, Inc. will do very well from a business perspective. Most of our users still do not know there is a company providing support subscriptions and the JBoss Network behind the product. The collective efforts of all of these vendors will help to raise the visibility of Professional Open Source in the market – you can see it already working with the broad proliferation of the term.
Can JBoss continue to gain market share?
The biggest macro impact of all of these announcements is the growing feeling that open source will become the majority of the market in many software segments. JBoss has already achieved a very large #1 market share (34% penetration according to BZ Research). With other vendors endorsing open source and having their own implementations, the overall percentage of the market that is open source will grow over the next 5 years. It is almost certain that open source will become the dominant (>50%) deployment platform for middleware in that timeframe, capturing all of the growth, and potentially some of the existing customer base of proprietary vendors.
This environment will enable JBoss to continue to expand our own market share for the base Application Server. Now that IBM, BEA and Sun have in effect endorsed an open source solution (and Sun has even endorsed the JBoss App Server now with our new partnership with the Sun Hardware group Certifying on JBoss), customers who previously would not evaluate an open source solution will now look at this path. When they look at the available solutions, the substantial technical and market lead that JBoss has accumulated will make JBoss a very likely choice. In addition, JBoss is offering our entire middleware stack as free and open source. So JBoss should capture a decent share of these new open source users coming into the market.
The JBoss strategy of mass adoption will make it difficult for any sort of significant penetration by other open source application servers. JBoss is free, it is high quality, there is excellent free documentation and wiki's, there is a wealth of free information available on the web and on the free development forums, and there are many high quality books and research done on JBoss. JBoss continues to stay in the forefront on important technology like EJB3 and AOP, while maintaining a very high degree of backward compatibility. JBoss is also very transparent - allowing anyone to look at our JIRA system to understand our roadmap and look for our time schedules for new features. JBoss has also attracted a very deep pool of outside developers who contribute code and their experience to the community. In addition, there is such a large ecosystem around JBoss - that allows users to buy other software that works very well and easily with JBoss. Even examples like working between JBoss and IBM MQSeries - there is a huge base of users and experience with this integration. So there is no compelling reason that will force an existing JBoss user away from JBoss.
The more interesting question that has emerged from this opening of Pandora's Box by these former foes of open source in middleware is that it begs the question of "If Professional Open Source is good enough for the Application Server, then how about for Caching, Portal, Integration and many other areas in middleware?" This will allow JBoss to expand even more into new middleware markets like Portal and Integration.
Is JBoss going to be able to continue to expand the community around the project?
The JBoss community has grown significantly since the emergence of Professional Open Source. There are a number of areas of growth:
- JBoss Employee Contributions. Of course the company has done well, so many former contributors to the projects now have real, paying jobs that allow them to develop on a full time basis for JBoss, Inc. This of course is the core of the company - people moving from contributors to making their living from this business. This is a lot of people writing free software for the rest of the community.
- Over 400,000 Developers now use JBoss. They contribute in many ways - from testing and finding bugs to creating added value components on top of JBoss.
- Over 5,000 Forum Contributors help each other to find the best way to use JBoss and come up with new and interesting ideas on how to improve the software.
- Over 500 Developer Contributors are signed up to participate in our developer forum and email list. Over time nearly 1,000 people have contributed code to the JBoss code base. Some are simple bug fixes, some are major pieces of functionality, some are great ideas for the next version, or even good ideas on what new projects JBoss should be creating.
- JBoss Open Source Federation is a mechanism that let's other open source project tie tightly to JBoss projects. It also provides a free hosting service.
- JBoss Local User Groups have just been started with our sponsorship. There are about a half dozen within two months of starting the program, and they provide a way to connect the JBoss community on a local basis.
- JBoss Customer Advisory Boards are a key way that we find out the needs of our customers. We have one for North America and recently held our first European CAB meeting.
- JBoss Network Subscribers are customers of JBoss, Inc., where we provide a mechanism to support, update, patch, distribute and manage large JBoss installations.
- JBoss Partner Advisory Board is made up of our top partners. They provide input to us for where their customers are going, and ideas on how JBoss can serve them. All of these partners also contribute in significant ways to the open source projects - donating code (such as Unisys with the Administration Console) or working on performance (like HP and Intel) or dedicating developers to JBoss projects (like Novell and CA). This helps to escalate the professional component of contribution.
As you can see, the JBoss community is very healthy. Our professional open source methodology is being spread out to many projects (23 total open source projects now get significant development effort from JBoss, Inc. - including Apache Tomcat and Eclipse WebTools projects). JBoss is a real community player.
Can JBoss replicate their success beyond the App Server?
Absolutely. We have developed a methodology around Professional Open Source that goes from the development to the community to how we do public roadmaps, to how we market and sell and partner. In December we introduced the JBoss Enterprise Middleware System - JEMS - as our overarching architecture and roadmap. Over the next 1-2 years we will build out an entire middleware stack. The unique attribute of this stack will be it is all plug and play. Meaning that you can use products stand alone - for example if you want to use Hibernate or JBoss jBPM with WebSphere - that is fine. If you don't need an app server, and just want to use JBoss Cache - that is fine. Of course we will make sure they all plug and play nicely together.
The JEMS stack is now what the community is talking more about than simply the application server. This does not mean we are turning our attention away from the app server. As you can see from our web site, we are actively building out JBoss 5 with a greatly enhanced Microkernel architecture. In fact this Microkernel architecture will enable easier plug and play between the different product sets.
We are approaching the market in a very deliberate manner by gaining success on individual project basis and maturing each project. We are seeing rapid adoption across our entire set of projects:
- Hibernate is now our second most popular project having risen from 20,000 downloads per month a year ago to over 80,000 per month on average the past 3 months.
- JBoss jBPM has risen from under 1,000 downloads per month before it became a JBoss project last October. Now it is well over 10,000 per month and rising.
- JBoss Portal 2.0 was announced in June as ready for Production Use. We have seen download rates on this rise quickly to 20,000 per month - outstripping all other large open source portal projects
- JBoss Eclipse IDE has not been downloaded over 600,000 times since it was introduced last year. This provides a nice set of add-ons to the base Eclipse package, and has some very cool new features for Hibernate and a JBoss jBPM design tool coming this fall.
The JBoss Portal release is a significant milestone for JBoss and the industry. Today it is still only 70-80% of the functionality of the larger Portal products. However, with the price of portals being close to $50,000 per cpu, a free and open source alternative can meet a great many customer's needs. We expect a number of our base application server customers that have portal needs to upgrade to this new product because of the high functionality and ease of development integration with things like Spring, JSF, and MyFaces. We have had customers benchmark the portal to great success - telling us that it is easier to install and use, runs faster and is more stable than other open source portal projects like Jetspeed and Pluto. We are also seeing partners jump into this arena with us. Most notably, Novell has dedicated engineering talent to work on the JBoss Portal, and will be donating dozens of their portlets to our open library. They are also supplying the WSRP implementation.Summary
In summary, the Professional Open Source model is gaining increasing momentum in the market - much like a snowball rolling downhill. JBoss has helped to pioneer this model, and the JBoss community is gaining tremendous benefits from it. A healthy and growing set of plug and play free and open source middleware projects backed by JBoss and our large partner ecosystem.