Another day, another survey – this one from Tools Vendor ZeroTurnaround. From what I can tell survey participants were self-selected – but the results underline what has been a solid trend over the last several years and I’ve seen the same in internal surveys I’ve commissioned.

Below is the 2009 / 2010 Container Popularity chart. Note the significant decline of Websphere and Weblogic and the growth in leaner, Open Source containers like JBoss, Jetty and Tomcat.

Glassfish bucked this trend – likely due to uncertainty about it’s future under it’s new owner Oracle. JBoss showed only a little growth – I’ll put this down to a fairly slow year in 2010. But 2011 is going to be very, very different. We already have a Java EE 6 Web Profile container (released last week) and JBoss AS 7 is taking shape pretty rapidly. With our increased attention to slimming the footprint and increasing the speed of adopting new technology and standards like Java EE 6 — my prediction is that JBoss will catch or overtake Tomcat in the next year.

So thanks for the opportunity to explain some of this.

First the obvious disclaimer – yes I work for Red Hat. Specifically I am the Director of Product Management for JBoss Enterprise Application Platforms and as such responsible for the product roadmap and technical direction of JBoss branded products like JBoss EWS, EAP and EWP.

So let me explain Red Hat’s model – something we call the Fedora / RHEL model internally. Red Hat provides subscriptions for use of its Enterprise distributions. A subscription provides the following (in no particular order) :

• long-term world-class technical support – and we do it very well (PDF report)
• long-term application compatibility
• long-term stability and predictability
• long-term partner certifications
• legal assurance
• long-term provision of security patches, performance enhancements bug fixes and RFEs

It may seem contrary if you’re used to the traditional model of “buying bits” but in our model, the provision of the bits is somewhat secondary; it’s something we have to do to support the value outlined above. For example, partners will only certify their applications and products if we have some way of identifying specific releases – supporting a continuous stream of releases is impractical. We can only provide application compatibility if we focus on specific identified releases.

So, one of the entitlements of a subscription is access to the supported binary distributions of a product – this is the thing to which we can apply all the other things I’ve outlined above.

For all of Red Hat’s products there are one or more upstream Open Source projects. In the case of JBoss EAP – the JBoss AS project is the primary components but JBoss EAP also includes Seam, mod_cluster, Apache CXF to name a few. Some of the projects that Red Hat uses in it commercial platforms are essentially Red Hat (or JBoss) projects – we provide the majority of developers, drive the roadmap and the release cadence (eg. JBoss AS, Seam, Hibernate), for others we’re merely one collaborator among many (eg. Apache CXF, OpenJDK, Apache HTTP).

The upstream Open Source projects is where the innovation happens – the focus of many of the Open Source projects driven by Red Hat is to act as technology incubators. Releases for projects like JBoss AS are frequent, experimental features are released, refined and re-released. That’s the focus – agility, speed, innovation. There’s never been any promise, implicit or otherwise that any given release is suitable for running your business critical applications. In fact we make it pretty clear on JBoss.org :

OK, so let’s dig into the relationship between project (or community) releases and platform releases. I’ll use JBoss AS (project) / JBoss EAP (platform) as examples as they are among the most widely downloaded / deployed :

Let’s take the JBoss AS 5 branch which was the foundation of the most recent JBoss EAP 5 family. JBoss AS 5 was focussed on a couple of big things : i) providing a new level of modularity via the Microcontainer 2.0; and ii) providing a Java EE 5 certified container. JBoss AS 5.0.1 was released in February 2009, followed a few months later by 5.1.0.

JBoss AS 5.1.0 met our functional criteria for JBoss EAP so that is what we picked up for our ‘productization’ process and JBoss AS 5.1.0 essentially became the Alpha Release for JBoss EAP 5.

The productization process is really not dissimilar to the kind of process you’d see in any other software company – we bring in all the major components, refine the dependencies, remove duplicates, perform additional testing above and beyond the community / project testing – focussing on security, performance, scalability, failure, longevity and the component integration points. We also look at documentation and the certification of third-party products like databases, Operating Systems, JVMs and other Application that work with JBoss. During this process we also run a traditional Early Access Program (aka Alpha, Beta) – this augments the attention the individual components receive during their own community release cycles. We’re fortunate to have some very willing customers who are able to apply significant resources to push our technology very hard using real-life applications and operational scenarios – often finding issues that are very hard to flush out in QE or during community release cycles.

The result of this process is an Enterprise Platform GA that differs from the upstream binary release we started with. First, we bundle additional components – like APR (Apache Portable run-time), Seam, mod_cluster, Apache CXF. And the core JBoss AS we include has a large number of fixes to address the security, performance and other issues identified during the productization process.

But that’s just the start.

JBoss EAP is supported for 7 years and with every additional minor or micro release we further improve the performance, security and stability of the Enterprise Platform. We’ve now released 2 micro and one minor release of JBoss EAP – that’s about 150 top-level issues in total. While the issue rate will slow over time – we’ll still be in a position to fix issues and respond to new security threats in 2016.

All those fixes are made available upstream and will ultimately make there way in to upstream binary releases but what the upstream project can’t guarantee is that those fixes will be isolated from more substantial changes and improvements – community releases typically don’t distinguish compatible bug fixes from more intrusive changes that provide the innovation.

OK so what happens to the community project once we’ve delivered an application platform? Well in the case of AS 5.0, from a Red Hat contributor perspective – the work was complete and Red Hat’s developers moved on to the next wave of innovation in AS 6 and AS 7. The goal of AS 6 is to deliver a Java EE 6 Web Profile implementation, the goal of AS 7 is to tackle the operational use-cases with a new domain model and console.

So to summarize this rather long post – if you want to deploy your business critical applications and receive long term support from Red Hat then the JBoss Enterprise Platforms are what I would recommend – if you’re more interested in seeing how those platforms will evolve and more interested in emerging technology but willing to take on more risk then upstream projects are where you should be looking. It all a matter of assessing the risk.

Replay Solutions just published a survey about Java Platform usage. The questioning, subsequent analysis and presentation of the results was poor IMO. But they did one thing right – they provided the raw data. Thanks for that.

There’s a posting on TSS entitled “Why is JBoss at the bottom of this list ???”. In typical TSS style – few people actually bothered to read the survey results or question them and a long and rather pointless thread ensues. This post is an expansion of my comment at the end of the TSS thread.

The original report has this chart :

Update – Sunday 4/11/10

My initial (very quick) analysis was wrong. My formula for searching for different categories had a basic reg-ex flaw so I was over-counting JBoss by fair bit. I’ve fixed that mistake (spreadsheet here) and also removed duplicates (responses with both “JB + TC” and “Tomcat” or “JBoss” – I’d already admitted to this minor double counting (in the comments) – the original author’s analysis still includes this error.

So the ranking is now the same as the original author’s but the %’s are different. My apologies to IBM for originally stealing their #2 spot By the way – Red Hat fully supports both Tomcat and JBoss AS – so #1 and #3 rankings and being able to satisfy 87% of the market isn’t such a bad result for us.

[As percentages of respondents - that's : Tomcat = 59%, Websphere = 39%, JBoss = 28%, Weblogic = 23%, Other = 19%]

These rankings (WAS above JBoss) are more representative of larger organizations (where both WAS and WLS have traditionally been stronger) – the latest Eclipse survey shows a similar break-out. Unlike the Eclipse survey – this survey doesn’t have any information on the respondents and they seem to be largely self-selected.

My original points still stand – poor questioning, poor analysis and presentation are common in these kinds of surveys. Always ask for the source data !

Open Choice isn’t some big, far-into-the-future vision thing it’s something we’re doing now. This year. Product-wise it consists of four offerings (where previously there was only one) and as we release them I’ll give you my perspective on why they’re important. Unfortunately I’m already a little behind – we’ve already delivered two products out of four and the third is in Alpha moving quickly towards Beta.

So let me use this post to talk about JBoss EWS 1.0 (Enterprise Web Server). EWS is basically a packaged, certified and tested bundle of Tomcat and Apache HTTP – the industry’s dominant Java web-container and Web Server respectively. We round out the bundle with mod_jk, APR and most importantly a management agent for JBoss ON. We currently support and certify on Red Hat Enterprise Linux and Solaris with Windows coming next.

Providing JBoss ON management support is pretty important – it gives customer the ability to manage the application and web stacks easily and consistently using the same toolset. If you want to learn more – there is a free Webinar on July 14th at 2pm Eastern – more here.

The rationale for supporting Tomcat is that it is absolutely the dominant Java web-container and has become an important part of the corporate IT fabric. Tomcat has been popular for years but in the last two or three  I’ve seen it evolve into a much more strategic platform for IT. Many customer I speak with have defined two distinct tiers of functionality – essentially a full Java EE stack and a lighter-weight Tomcat platform. By supporting both the dominant Java EE implementation (JBoss EAP) and Tomcat, combined with the ability to manage from a single tool – I think we can do a much better job of satisfying a much broader customer base than our competition. Here’s an (albeit unscientific) chart from a recent survey that demonstrates this well :

It’s also interesting to note that from this survey JBoss’ deployment share is more than Websphere and Weblogic combined.

OK, so JBoss EWS 1.0 is out and we have customers deployed or deploying some pretty large, strategic apps. but I’m already thinking about the next version (code named Cavalier). Some initial ideas for Cavalier are :

• increasing platform support to include AIX, HP-UX and maybe other Linux flavours;
• alternative Connection Pool implementations for Tomcat;
• looking at a more recent version of Apache HTTPD;
• possibly supporting mod_cluster.
• soft-appliances to better support virtualized hosts.

Any other thoughts are always welcome – leave a comment or get in touch directly.

OpenLogic have the results of a survey on Open Source web / application servers. The only surprise is Spring’s DM Server appearing so quickly – I actually thought it might take longer to get traction given the amount of competition in this space. Glassfish is also making good progress.

37% Tomcat
26% JBoss
18% GlassFish
4% Geronimo
3% Caucho Resin
3% SpringSource DM
1% Jetty
0% JOnAS
7% Other

Compare the competition and acivity with the Proprietary space – just 2 vendors trying to maintain their relevance and support revenue stream. Interestingly – the major impediment to adoption cited in the survey is “Robustness/scalability/performance”. Interesting because that’s often one of the reasons many of our reference customers move to JBoss. I’ll write a fuller response to this survey next week when I’m out of vacation mode.

I’ve lost track of how many “predictions for 2009″ I’ve read but they nearly all claim that the world-wide recesion will be a catalyst for open source adoption; many of them mention JBoss as the poster child for Open Source displacing encumbent, proprietary alternatives. Here’s one such quote from Dr. Dobb’s Roger Burkhardt :

… traditional companies like Oracle to actually raise prices during the economic downturn. The company’s 45 percent increase on its BEA acquired-WebLogic application server, for instance, is causing customer migration to JBoss’ open source enterprise application server“

Finally – I know it’s some time off but you might want to put JBoss World 2009 in your calendat – it’s in Chicago on September 1-3rd. And next month we have the JBoss Virtual Experience – an online conference for JBoss Developers, Customers, Users.

I promised myself I wouldn’t bite; yet here I am. I’m contributing to the “Here’s what Sun should do meme”. And why not -  I’m as good a keyboard-CEO as the next person and I did work for Sun for almost 9 years.

But first I should note – I think Tim Bray was pretty bold starting this (as he’s still a Sun Employee); and Sun is a pretty unique company in that it’s OK to do what Tim did (assuming Tim didn’t get booted).

#1 Java. I’m part of the Java ecosystem so it’s important to me. Basically, do what James says:

…, at worst go back to SUNW as a stock ticker.”

then do what Tim says :

As for Sun’s role as Steward of Java, and in particular the Java Community Process, let it go already. Java has mostly won and is mostly the establishment, and the community is smart and conservative enough to keep anyone from doing what Microsoft tried last millennium, or in any other way to subvert Java’s interoperability. In 2008, the JCP is costing Sun opportunities and friends and gaining us very little that I can see.

So I’d like Sun to set the JCP free, turn it over to the community, and when we develop some cool Java-based technology in-house, take it to market, try to make some money with it, and after it’s caught on and the bugs are shaken out, consider whether or not it ought to be taken to the JCP.”

But the community is going to have to continue to be smart and I also think that a consortium of interested parties would have to step up and invest some resources in making it work really well. Getting the balance of commercial interest and community is the key – I think Eclipse has it about right. Money makes the world go round – sorry if that offends anyone. Oh and while we’re at it – OpenJDK should be treated like the Linux Kernel – we only need one code base (note IBM and Oracle).

#2 NetBeans

I’ve used both Net Beans and Eclipse (not for a year or two mind) – Net Beans is better, more polished, etc. But it doesn’t matter – it’s not a sustainable product. When Sun stops it’s investment – it will fade away. Eclipse won in mind share years ago; Sun should have recognized this in 2005 – I did. Java needed something like NetBeans in 1999 to attract developers but not anymore.

#3 Solaris

See above. Linux has won. Whatever technical merit Solaris has today will be commoditized next year. The innovation around Linux is relentless.

#4 Middleware

Sun has some great products and technology; some that are popular and fit well with Sun’s demand driven model (eg. Glassfish) and some that make money (Identity Management). Figure out what you need to do #5 – choose the best available OSS solution and run with it.

#5 MySQL

Invest, Innovate. Hire sales people who can sell data-based solutions – steal $1bn from Oracle this year, $2bn next year, repeat.

Demand for storing, archiving, cleaning, replicating and accessing data is going to grow. Forever.

#6 Storage, SPARC, Volume Servers

Keep only what is required for #5.

#7 Open Office

I’m a beneficiary of Open Office – I use it exclusively but have never paid a penny. That’s the problem – I use it because it’s free and it’s the best Office Suite available for Linux. There is no business case here. Thanks for a great product though.

#8 Java FX

Drop it. Nobody will notice.

If this sounds depressing – I’m sorry but it should be clear that Sun isn’t going to be the company it was in 1999. It needs to adjust its cost-model inline with its Open Source strategy; it needs to be much smaller; and needs to do much, much less, much, much more sucessfuly. Even if Sun goes private.

IBM’s WebSphere application server is now ten years old, and during that time, it has evolved and matured into what its users think of as the best application server anywhere, but most especially in the large enterprise market where IBM has traditionally had its home. Make no mistake about it, WebSphere is a powerhouse in many ways, and its users truly love this
product.

Fantastic – already written in press-ready language – could be lifted and dropped straight into a press release. So good – it could have been written by Websphere’s marketing team themselves !

OK, by now you think you’ve seen through by thinly veiled blast at EDC’s report. You think I’m bitter because JBoss came in 5th (out of eight). Right ?

Well hold on. Though I have some issues with the report which I’ll get to – JBoss actually did very well – after all we beat Weblogic – which is no small feat. And there are other bits of the report worth highlighting. Here’s one quote I like which enforces what many other analysts are saying about JBoss and something that differentiates us from our Open Source brethren :

JBoss Enterprise Application Platform competes with Oracle’s Application Server, WebLogic, and IBM’s WebSphere in the high-end market for large corporate applications. The recent acquisition of BEA by Oracle may provoke some consternation and uncertainty amongst the end users of both WebLogic and Oracle AS, which in turn provides an opportunity for both JBoss and IBM.

And another that demonstrates that JBoss is focussed on our customer’s highest priorities :

JBoss really shone in the areas of security where users gave it the best ratings of any product in the survey, compatibility with other software, and the very important value to cost ratio.

The second point is worth repeating, given the economic situation we find ourselves in – JBoss had the best value to cost ratio.

So I’m actually not bitter – if you’re the underdog competing with some significantly larger competitors – this kind of validation is golden and It shows that JBoss is till punching way above it’s weight. I’m not bitter but my spidey-senses are tingling; there’s something just not right about this report.

Methodoloy

I find EDC’s methodology odd. It’s not clear how many responses were received for each question or for each vendor. For example you would expect to see many responses for Windows Server 2003 vs something like NetWeaver (which isn’t quite as main stream) – the number of responses is significant both statistically and as a proxy for adoption / acceptance.

Bucks the general consensus

The results just don’t seem right. I’ve worked for 2 of the vendors on the list (previously Sun and currently Red Hat) and know the other vendors really well having been involved in this space for as long as anyone. Though I trust my instinct – that isn’t good enough so let’s compare some other data. Forrester recently released a report (also based on a user survey) covering some aspects of the EDC report.  In that report WAS 6 did significantly better than WAS 5  but generally scored lower than JBoss. For example :

The EDC report doesn’t correlate with the summary points of the recent Burton report on JBoss. Also it doesn’t correlate with our own (sponsored) satisfaction surveys (like this). I find it strange that WAS scores so well and Weblogic scores so badly – putting them in seventh plance out of eight just seems a little too extreme and a little to convenient. I’ve yet to personally meet a Websphere customer who says good things about WAS – maybe I only meet the customers who’ve already decided to move to JBoss ?

Third – I talk to customers every week and Red Hat’s sales team a couple of times a day and I just don’t see Geronimo mentioned at all – few if any user satisfaction surveys actually call out Geronimo – what was it about these respondents that give them an unrepresentative affinity towards Geronimo ? I just don’t understand why Geronimo is in the survey.

So a couple of request to EDC – 1. open up the unfettered results; 2. provide a little more detail on the methodology; and 3.  confirm that this was a purely independent survey that wasn’t paid for or unduly influenced by any of the vendors included in the survey.

It’s great to see the innovators rallying around such a good cause – but look at the dates !

“provide an up-to-date server model that gives enterprises the features they want.”
- Rod Johnson, Spring Source CEO – April, 2008 [link]

“GlassFish v3 has a modular, lightweight, extensible architecture”
Sun Employee, glassfish.net – June, 2008 [link]

“JBoss’ modular architecture means customers can also choose the features they want, instead of installing a full J2EE application server on devices with limited processing and memory capability.”

Marc Fleury, JBoss CEO, December, 2002 [link]