Rise or fall for Red Hat?

An analyst friend emailed me the other day to get my opinion on the analyst community's negative Red Hat pile-on. Bank of America, Global Equities Research, and others have recently been hand-wringing over Red Hat's future, suggesting that its JBoss business is stalling, that it's "losing momentum" in emerging markets like China, and (here's the one I find immensely laughable) that hardware vendors like HP are having to step in to fill Red Hat's failing shoes on support.

Trip Chowdhry of Global Equities Research tried to outdo this list, however, arguing that not only does Red Hat stink, but the entire LAMP ecosystem is rubbish, too. He declares that .Net is winning developers' hearts and minds while the LAMP stack is on its way to relegation to the dustbin of history. His bias exposed (just as mine is here: I'm an open-source believer and see rampant uptake of LAMP and open source throughout the enterprise), he has his 15 seconds of fame. Time to move on.

One thing, however, bears further investigation. JBoss. The consensus view from analysts seems to be that JBoss is Red Hat's Achilles Heel. Let's take a look at some data to get a better picture.

1. While JBoss isn't infallible, it clearly is the leading open-source application server. BEA as a standalone vendor doesn't have much of a future. Only IBM is a truly credible long-term competitor from the proprietary camp.



JBoss vs. Geronimo
(Credit: Google Trends)
2. That said, what about competition from open-source application servers? Looking at the Google Trends comparison between Geronimo and JBoss, it would seem that interest in JBoss has declined steadily since 2004 while interest in Geronimo has remained steady (with a bit of a rise, though not much). Looking at news volume, JBoss seems to have flattened out in the past year.
   
   Against this data, however, consider the following:
   
   
   • In the time that interest in JBoss was supposedly waning (as measured by Google searches), JBoss revenue was exploding (as measured by ever increasing numbers of enterprises happily throwing money at the company, with revenue growth shown at the bottom of the chart below), even as downloads continued to rise. It's hard to make a compelling argument that interest in JBoss was slowing as downloads and revenue increased.
   
   
   
   JBoss Downloads and Sales Growth
   (Credit: Matt Asay)
   • Want to know where most of the Google Trends Geronimo interest is? The Philippines, that source of untold billions in IT spending. Second on the list? Italy. Third? Argentina. Where was JBoss popular? In countries with a bit more cash to spend: United States, Belgium, Canada, France, Spain, etc.
   
   In other words, a picture may be worth a thousand words, but in this case it's not worth a thousand customers.


3. Even so, I suspect that there has been developer movement away from JBoss with Red Hat's transition of JBoss to its "Fedora" development/release model. I am sure there was plenty of developer angst about the move, just as there was when Red Hat originally split its Linux efforts into Fedora and Red Hat Enterprise Linux. If memory serves me well, however, this momentary frustration has not left Red Hat financially crippled. On the contrary, it weathered the initial storm and has ended up being far better off for the move.


4. More pertinently, I believe JBoss' biggest competitor is not Geronimo and in the short term it's not really IBM's Websphere or BEA's Weblogic. It's Tomcat (which, ironically, is also developed by JBoss).
   
   
   

   Application Servers Used with Alfresco

   (Credit: Alfresco)
   It would be convenient to think that Tomcat is not used for any serious applications. It would also be wrong. While just one data point, Tomcat is at the heart of a majority of Alfresco deployments. I don't have the data to know what percentage of these are mission-critical deployments, but I can tell you that Alfresco's open-source web and collaborative content management solutions are used in a wide array of mission-critical applications. Tomcat is the application server of choice for a significant percentage of these (of which I'm aware).
   
   We always push enterprises to use supported software because it's the wise thing to do. Enterprises still deploy a heck of a lot of Tomcat. This, of course, is a latent opportunity for Red Hat, but for now it's a potentially pesky one for the company. Even so, we're doing an increasingly large amount of business with the JBoss team. I doubt we're alone.

In short, I suspect that JBoss has suffered a bit in its business model transition, and it may well be the victim of its own success with Tomcat. But these are temporary setbacks that can fuel long-term growth.

The analysts are stampeding toward the exit, mindlessly bleating the same message of alarm. While Red Hat clearly can do better digesting the JBoss acquisition, I've seen nothing to suggest JBoss' best is behind it. For Red Hat and JBoss, the future is wide open. It just needs to continue to execute well as it generally has.

[royrusso]

Analysts aren't looking at 'Google trends' data. They're looking at the amount of money RHT spent on the JBoss acquisition and the poor revenue return on it. That sorta tends to happen when you destroy the sales and marketing machine JBoss had built.

I value your opinion on everything (including pie-baking), but you need to stop looking for blame in the shadows and bushes. The fault lies squarely on the shoulders of the RHT management team for rolling ~$300m down a rat hill.

Then again, maybe RHT bought JBoss so ORA/IBM wouldn't get there hands on it, but I can't give RHT management that much credit either.

[Matt Asay]

And you need to remember that your perspective is soured by you and the other JBossers walking in and expecting to dictate the show. There are plenty of quality JBossers left. Shaun, for one, seems to be doing well there.

Btw, when have you ever known a sales and marketing team to be kept largely intact post acquisition? I think the JBossers thought they were special somehow; that the rules of M&A wouldn't apply to them. Surprise!

I'm not looking for any blame. I've yet to see any reason to believe that Red Hat isn't getting JBoss moving. I know there are plenty of JBossers who have left the company who are happy to dig at Red Hat not loving them enough, but I'm not overly bothered by their outside perspective on what's going on inside the company.

Regardless, you of all people should be hoping that Red Hat pulls it off and should be doing your best to help make it happen. Have you no pride of ownership?

[royrusso]

"Btw, when have you ever known a sales and marketing team to be kept largely intact post acquisition?"

Its common to see sales and marketing teams merged. The mass exodus that occurred THAT EARLY after the acquisition isn't normal, sorry Matt, and it had a negative effect obviously on earnings for the division.

"I think the JBossers thought they were special somehow; that the rules of M&A wouldn't apply to them. Surprise!"

Remember... what was RHT buying? They weren't buying code (OSS). So they were buying... yep... people! As luck would have it, once the "special" people started leaving, the investment's downward spiral began. Today it seems the surprise is on RHT.

"Regardless, you of all people should be hoping that Red Hat pulls it off and should be doing your best to help make it happen. Have you no pride of ownership?"

So consider it a "get your sh*t together, RHT!" post. Would you rather I send a private email to Szulik? RHT didn't listen to me when I was working there, why would they listen to me now anyway? ;-)

[lemmingme]

Mr. Asay, I wanted to add my insiders comment to your posting (do i work in RH? Maybe, maybe not - does not matter)

Your diehard support of RedHat/JBoss is admirable, but hardly lends credibility to your supposed stance as an industry, neutral observer. Your allegiances are clouding your subjective side.

Let's start analyzing your blog:

1 - JBoss As is the leading AS offering. True, but only by a much shorter gap than you can possibly fathom or understand. You discount BEA much to easily due to your blinders on that JBoss is better. That is just not true, there are a lot more, high end mission critical deployments which use BEA vs. JBoss. Have a look at BEA's revenue, in decline as they are, are still much much more then JBoss' - and this is just looking at the BEA App. server licenses sold today. Again, I know they are in decline and yes BEA is hurting, and will likely die a slow death unless they let Larry have the body now before it rots.

2. Competition. Several factors have come to play against JBoss AS. First and foremost, their lack of delivery on AS 5.0 is hurting us/them. Vey simply, Geronimo, and even GlassFish now (which you conveniently forgot to mention), and of course Tomcat are now eating into the revenue pie which almost exclusively for JBoss only 2 years ago for a support offering. Another factor at play, is perceived lack of the need for paid support for a product, which is much more widely understood today, and the community support is actually quite good compared to 2004/2005 timeframe.

3. Fedora style splt. This is porobably one of the major reasons why even internally within the JBoss teams, people have stopped believing in the little engine that could. If you had been around in the year before RH came into the picture, you would have heard guys like Fleury and Sabrin talking about how we would never do this, how this would go against the very core of what open source and open access means to the community - the very people who helped make JBoss what it is today. Linux weathering the storm as you put it, serves as no basis for assuming that the app. server community would react the same. Have you been reading the same manual that the RH exec's have been feeding themselves for the last several years? Splitting the App. Server a la Fedroa was not a good move.

Some general comments about your reply to Roy Russo now.

1 - I must say, firstly, that you should not read offence into this, but let's be real here. Alfresco, is hardly a "mission critical" piece of software. It's a content management framework, built on top of some open source components. A transaction processing system, a claims system, etc. are all mission critical pieces of software. Do some people use Alfresco as a mission critical platform, I would guess some do, but then again, people also build homes right along the eastern seaboard then complain when a hurricane comes and wipes it all away. Again, this is not a personal attack, I am just stating that App. Servers and the applications that a company uses to run it's enterprise systems are a little more mission critical then a content management portal.

2 - You don't truly understand the magnitude of the change that came about when the RH policy driven exec's with their shiny MBA's came in and ordered a wholesale gutting of the JBoss sales organization. Why did they do it? Ask them now, and see if they agree it was a wise move. The top sales exec's all left JBoss - there are maybe 2 or 3 left from the heyday of JBoss subscriptions and the all to present "Viva la sales rep." email message that would be sent out by SalesForce everytime a new subscription deal was done. Those messages were flying in a lot every day. Now? They even turned this feature off in the quest for "corporate governance" and "insider trading fears" that is truly the hallmark trait of RH Culture.

RH is more concerned with creating policies and initiatives that ran contrary to the spirit which Fleury and his team created - one of openness, caring for their work, and the motto have fun which was present every day till the day RH took over. RH was more concerned with making sure no one on @core used profanity, or communicated anything with anyone about any of the projects we were working on. It became the Borg Collective, and that hardly inspires creative and talented people to keep on doing what they do best.

You should do well to listen more to the so called 'outsiders' who you dismiss as knowing nothing about what goes on inside of JBoss. You would be surprised just how many of these outsiders still have long lasting friendships inside of JBoss, and those friends tell them of all the amazingly silly things management does. You were never a JBoss employee, how can you truly know what it was like, or what it is like today? Listen more to the ex-wives club, you might learn more then you think.

Did JBoss'ers think they were special, sure we did, and we were. We were the trailblazers, doing something innovative, and different. We fought the likes of IBM, BEA who tried to drown and kill us at every turn, and we survived, and even managed to come out on top in many ways. So yes, we are a special bunch, and proud of it. It's to bad RH management tried to take that feeling away. They are paying the consequences for it now, and there will be more to come. There are rampant rumors within the organization of more departures to come, which only gets worst as the stock price stagnates, and those $1.xx something options vest out every month.

Lastly, as annoying as Roy can be, he has more insight and credibility talking about JBoss - he was there, he saw it happen. Who would you listen to, Roy, or someone who uses Google Trends data?

Peace

[royrusso]

"Lastly, as annoying as Roy can be..."

This is clearly someone that knows me well.

[Matt Asay]

Keep in mind that I'm on Roy's board of advisors for Loopfuse, so I know him reasonably well. I also have a range of ex-JBossers working with me and keep in close contact with Bearden, Fleury, and others. So I think I have a reasonably good handle on what the insiders and outsiders at JBoss think.

I'm not saying Red Hat has done a stellar job of integrating JBoss. I think you'll have a hard time finding that in my post. Quite the opposite. I just said that I think Red Hat will figure it out.

As for Alfresco, you're right, to a point. But I think mostly because you're thinking of ECM as it was, rather than as it is being used now. Content is (or can be) at the heart of the most critical applications a business has. Consider what happens to Expedia if it's website goes down because its CMS dies. Or consider what happens if Merrill Lynch's trading system goes down. Etc. These are all content-based systems, and we're running the infrastructure of their competitors (sorry, I can't name names on this).

As for the Fedora move, I think you may be overstating your case. The original JBoss team had a decent services model. But I sold alongside JBoss and watched the frustration (and panic, at times) when companies standardized on JBoss and didn't pay a dime for support. Then along came the JBoss Operational Network and JBoss learned that the ideal of giving everything away was a great theory but not so great in practice. You have to close off some service/some value or you don't have a business. I strongly prefer to keep everything open but if you don't have some analog value when you're giving away digital code, you're hosed.

[jharrop2]

Hi Matt, you said "Tomcat (which, ironically, is also developed by JBoss)"

Tomcat is of course an Apache project ... http://tomcat.apache.org/

cheers, Jason

[acoliver]

Tomcat vs JBoss is a little like "gee tires are beating cars because they outsell them more than 4 to one". That trend was ALWAYS there. Your analysis is flawed.

[mraible]

> Tomcat vs JBoss is a little like "gee tires are beating cars because they outsell them more than 4 to one".

I disagree. Tomcat is perfectly capable of deploying highly scalable applications on. JBoss and Java EE simply isn't needed (as much) now that we have Spring. ;-)

[acoliver]

mraible that is moronic. Java EE is needed to adequately power spring or more aptly a transaction manager is needed. You miss the point, tomcat is part of multiple appservers. It is a different market.