Sunday, August 19, 2012

JBoss Recollections Part 3 - The Tech Team

Check out Part 1 and Part 2 before reading this, or you may get lost and confused.  Part 3 involves my memories of the Technical Team, which was the heart and soul of JBoss.  The foundation that made everything possible.  BTW, in this article, it is worth clicking on the links - there are some historical and hysterical ones...

The chart at the right is from June, 2004.  As mentioned earlier, I did not start working with JBoss until September, 2002.  I won't go into all the folks on the right, and I'll add a few.

The one important person involved with the JBoss technology that I never met was Rickard Oberg.  Many give him credit for the Microkernel architecture and building much of the early versions of JBoss.  He left before I got there.  I was never clear on what happened, but my impression is that both he and Marc had strong opinions and that finally led to his departure.  I do know that Marc always talked well about him.  Marc offered him some of the Economic Interest Units in the Spring/Summer of 2003 even though he was no longer doing work on the project and there was a deep rift.  Rickard refused them, and I am not sure he is really the kind to regret the amount of money he would have made if he had simply signed the document.  Rickard often criticized JBoss in later years - just google rickard oberg jboss.  As typical with many things JBoss, a lot of emotion.  As Rickard left, the guys who became the "real" core became increasingly involved:

Scott Stark.  I mentioned him in Part 1.  Marc viewed Scott as his real partner.  The above chart has Scott at the top, but every developer at JBoss would have put him there by his shear force of contributions and leadership by example.  Scott lived out in Washington and would work crazy hours.  The only time he was not working was if there was fresh powder on the ski slopes he lived next to.  Scott made sure that JBoss builds actually worked for users.  While Rickard may get the credit for the Microkernel, Scott gets the credit for making the thing work - and work well.

Sacha Labourey.  Sacha is not shown in this chart because he was running JBoss Europe by this point.  His initial contribution to JBoss was writing the Clustering.  This was the beginning of bringing JBoss from just a developer tool to being used for real project deployments.  He and Marc built a very strong relationship.  Sacha headed up the creation of the European operation, including sales and marketing.  But he always had a hand on the technology.  He earned such respect and recruited people so well about 15 of the world's best middleware developers wound up moving to Neuchatel, Switzerland where Sacha lives.  Sacha became the CTO when JBoss was acquired by Red Hat and can be credited with holding the technical team together thru a pretty messy acquisition.  Sacha is now the founder and CEO of CloudBees, where I get to work with him still (lucky me!).

Bill Burke.  Bill was probably best known for implementing EJB in JBoss.  That is a bit of an understatement as he had a wide influence across a lot of things at JBoss and along with Scott was one of the key guys that made sure things actually worked.  It is also a bit unfair as EJB has taken a few shots from later OR mapping like Hibernate, and Rod Johnson's Spring pitch.  Bill was not wedded to the worst of EJB, but the best of it.  I think he was a pretty big part of getting Gavin to come to JBoss as an example.  As a Patriots fan from Boston, the only thing that bothered him more than a loss to the Giants was when Spring was trying to take credit for Hibernate ;-)

Adrian Brock.  I first got introduced to Adrian via email from Marc saying he headed up Support.  I later learned that Adrian had started answering questions on the forums.  He is one of those rare, brilliant developers who can read code like speed readers go thru a book and understand and remember everything.  He had a RTFM attitude (click on his blog link above to get an idea) on the forums, but if the question was valid, there was not a better person to help find the answer or to create a new feature.  Here is a great example of how Marc got people like Adrian involved in the project.

Bela Ban.  I remember riding in a taxi on 101 near the SFO airport with Marc while he was on his cell phone trying to convince Bela Ban to join JBoss (maybe early 2003).  I didn't know Bela then other than he was the brilliant developer behind JGroups.  I think he was evaluating whether to join Google or JBoss or some other company that I forget.  I've had a number of great runs over the years with Bela - once running from the Marriott in San Francisco over the Golden Gate Bridge to Sausalito.  He had left his car there and done the reverse run the day before.  He has also done the Jungfrau Marathon. And he is still going strong on JGroups - I just saw him at JBoss World a couple of months ago.

Gavin King.  Well, my favorite Gavin King story is when he came to give a talk at Villanova for a JUG.  My 70+ year old non-programmer father happened to be in town and had recently read Java for Dummies.  I introduced him to Gavin describing how he had changed how Java developers work with databases.  My father misunderstood his name and then called him "Kevin" every other sentence as he questioned him about everything he had read in the Dummies book for the next 10 minutes.  And I did not interfere - it was far too funny.  Anyway, Gavin is the creator of Hibernate as well as a number of other technologies.  It is safe to say that neither JBoss or Spring would have been anywhere near as successful as they were had Gavin not done Hibernate.  He and Marc had a love-hate relationship the entire time.  Thank God for Ben Sabrin, peace keeper extraordinaire...

There are so many great people on this chart.  Remy Maucherat  was a key committer to Tomcat for years, and a very important part of JBoss.  Andy Oliver became a very important face for JBoss to many customers over the years.  He now had his own business continuing on a path he successfully pioneered at JBoss.  Julien Viet and Roy Russo created JBoss Portal with just two people.  Tom Elrod and Roy went on to found Loopfuse several years ago.  This has gotten long enough - so apologies to all the other fantastic contributors!

The amazing thing about the team was how distributed, yet how tight it was.  I was reminded that the quality of the developer far outweighed any management or development technique.  Forget Agile - just get developers like this crew and you will have a successful project.

If you want to read a good article about the early days of JBoss development and some of the key guys - take a look at http://sourceforge.net/potm/potm-2003-04.php.

Tuesday, August 14, 2012

JBoss Recollections - Part 2

This is Part 2 of my personal recollections of JBoss.  Please read Part 1 first, unless you are one of those people who eat desert first.

In Part 1, I had left that first meeting in New York being very impressed with JBoss and Marc.  We agreed to meet on October 7, 2002 in Atlanta.  We actually met in my hotel room with Marc, Ben and Marc's father, Daniel.

Daniel was an elegant man.  I know that may sound strange - but between his French accent, his strong experience of the world (He had been an executive for P&G in Europe), his sharp questions and easy manner - I really liked him.  Marc credits him with a lot of the inspiration and confidence to do JBoss.  You can see a dedication and picture of Marc and Daniel on the right in Marc's "White" paper.


We had a number of discussions between our initial meeting and reviewing this set of slides that I have sprinkled on the side of the page here.  In this meeting we kind of laid out the basic roadmap for the company for the next several years.
Simple Plan - You can see there is a lot of pretty simple stuff in here.  Take our time and work our way up the ladder.  Learn from Linux and Red Hat (funny Red Hat was just turning profitable back then).  Partners.  Press.  Move from Developers to Deployment.  Move to a Production Support model where there were multiple points of value.  It took us a while to get to this real vision - about 2 years before JBoss Operations Network debuted as an example.
Revenue - was from Training, Documentation Sales and Consulting/Support.  I think 2002 ended around $1M.  Ben and Marc had come up with a clever way to sell support in those early days - basically sell a minimum of 50 hours of phone support.  If a customer used only a couple of hours, then there was great profitability in that - and that was the normal pattern.  This allowed a price tag of like $5K, which was high enough to support an inside sales model.  During 2003 Ben hired Mark Andres as an inside sales rep - he was just out of school and turned out to be a great part of the team.  We ended 2003 with about $3M of revenue.

PR - During the end of 2002 and into 2003, JBoss hired a PR firm to start giving it more visibility.  The company was a hot topic.  Marc was very charismatic and gave great one liners that the press loved to pick up on.  We picked fights with the big guys and everyone loved the David vs. Goliath story.  In later years we would time a quarterly press release that went to every financial analyst who covered BEA the day before BEA announced their earnings.  So when Alfred held his quarterly earnings call someone would always ask him about JBoss and he would blow up.  Very funny!

Tech Differentiation - While the business side was trying to figure out the direction, the open source project continued to execute.  The key feature in JBoss is not talked about that much.  But basically a developer did not need to recompile their whole project every time they made a change like they did in WebLogic and WebSphere.  This meant the 100 changes a developer made in a day were immediately executed rather than wait 2-5 minutes for a build in another system.  Many, many WebLogic and WebSphere shops were using JBoss on the developer's desktops.


Economic Interest Units - In my next post I am going to talk about the technical team.  At this time, there were only about 6 real employees at the company.  But the network extended beyond this.  Marc had been creative in terms of trying to farm out some consulting and training work to a lot of them to keep them interested in working on the project.  In 2003 Marc did one of the most amazing things I have ever witnessed - he simply gave away "shares" in his company to probably 30+ people who had made contributions in some way to JBoss.  This act of generosity and recognition that there are a lot of people that deserve credit in the success of a company was impressive and earned my deep respect.  Since it was still an LLC, he came up with the idea of Economic Interest Units.  These later became shares when JBoss became a Corporation.

Jim Green - Jim Green was the CEO/Founder of Active Software, which was acquired by webMethods.  He was the first strategic user of JBoss that saw the potential and became an important advisor to Marc and the company.  Their endorsement of JBoss was a critical step in making JBoss "acceptable" for beyond development.

Ukiah Retreat - In the summer of 2003, Marc assembled a group that included Sacha Labourey, Jim Green, Larry Augustine and myself along with our host, Larry Rosen.  Larry has a beautiful house outside of San Francisco in Ukiah.  We met for a day to discuss the future of JBoss.  Many of the topics brought up in the slides above were discussed.  The Sun J2EE friction was discussed and I think we made the decision to finally bite the bullet and pay Sun their extortion fee for saying we did J2EE.  It was also at that retreat that I started thinking about JBoss on a more full time basis, which happened in September of 2003.

In Part 3 I am going to talk about the technical team...

Sunday, August 12, 2012

JBoss Recollections - Part 1

I just read Matt Asay's article - Unstoppable JBoss 'mafia' has big tech biz in its crosshairs.  It really got me thinking about the old days and all the great people at JBoss.  I have no idea if my motivation (and available time) will enable me to finish, but I thought I would share my recollections and some of the stories I saw unfold at JBoss.  While I actually never really did anything at JBoss, I was in a position to see a lot of things happen across all aspects of the company.  Hopefully I won't mis-represent anything, and encourage others to correct me, or add to the story lines, or create their own recollections.  It was an interesting ride...

Meeting JBoss
My first interaction with JBoss was shortly after Bluestone had been acquired by HP.  We had become interested in the idea of open source for our app server, but were precluded from it by our licenses with Sun for J2EE.  Rich Friedman was then the CTO for us and he tried to make contact with JBoss to see if there was something we could do together.  JBoss was (smartly) not interested at the time.  I forget exactly, but the rejection from Marc Fleury had the typical spice to it, and I remember laughing with Rich about it.

In 2002, HP had acquired Compaq and we were shutting down the Bluestone/Middleware division.  Marc sent a letter to myself and Carly in August, 2002 (see a copy on the side).  I called him back and told him I doubted if Carly would respond since she had decided to pay BEA $5M to transfer the Bluestone customers to WebLogic, but that I was going to be free shortly.

I went to New York City in September, 2002 and met with Marc and Ben Sabrin for the first time.  Marc was giving a training class in a hotel there.  And I got to hear the amazing story for the first time.  Some of the items below I probably learned later, but I will lump them in together here to make this series somewhat chronological.  Some of this is probably wrong, but it is my recollection of the stories.

The Start of JBoss
Marc worked at Sun as an evangelist and got tied into the J2EE efforts.  He got to know Vlada Matina and Mark Hapner, two of the key architects behind enterprise Java and EJB.  Marc become very interested and started up an open source project called EJB-OSS.  It got some early traction and some others started joining as contributors to the project.

His first attempt to make a living out of it was an attempt to create a company that would basically put up a J2EE Platform as a Service.  Of course that was before the Cloud was cool, and the VC's did not invest.  Marc loved to tell the story of how Doug Leone of Sequoia said "This is not a bad business plan, this is a HORRRRRRIBLE business plan."

When the money ran out, he and Natalie moved from California back in with her parents in Atlanta.  Marc worked in the garage, literally.  Another great line we used to say was "Marc didn't start this business in his garage - he started it in his In-Laws garage."

Training!
A few users were asking for help and training.  He decided to see if anyone would be interested in a training course.  He posted it and filled a class of 20 for $3,000 each.  He and Natalie did a few of these - taking trips to places like Australia, renting a hotel conference room and collecting $60K for a week's work.  All of a sudden, he was in business!

It also turned out that a number of the contributors to to the project also wanted to get some side work in - so Marc started farming out some of the training.  Juha Lindfors picked up on this and eventually led the training efforts - creating better and better material and delivering a number of the courses.  This also led to more dedicated open source contributors, which of course had some great effects on the project becoming better and better.

Scott Stark & Documentation
One of the best things to ever happen to the project was a contributor named Scott Stark.  He was a Bear Stearns developer, specializing in security I think.  One of his most important early contributions was to write some documentation.  I forget the early arrangement, but I think he got a % of all the documentation sales.  Documentation was an important early revenue driver.  My memory is fuzzy, but I think we were doing about $20,000 per month in documentation sales by the 2003 timeframe.  More importantly he became the real rock of the development team - putting together all the builds and making sure JBoss was a quality project.

"Doesn't Suck"
Ben Sabrin had a unique blend of talent as a recruiter in Atlanta.  He has the sales person's natural talent to talk with anyone about anything.  But more uniquely, he could talk with developers and really understand what they were saying.  He also had a nose for new trends, and he was seeing the rise of J2EE and this little upstart JBoss.  Probably in the 2001 timeframe, before he joined JBoss, he organized a session for developers where each of the J2EE vendors would speak.  IBM and BEA reps stood up and gave their normal boring presentations.  Marc arrived late without a laptop and saw others were giving presentations.  He borrowed someone's laptop and put together his two slides.  First slide: "JBoss is Free" - great applause.  Second Slide: "It Doesn't Suck" - brought down the house.

Well, my first meeting with Marc and Ben was a lot of fun.  They were clearly riding a huge and growing open source project just at the right time.  It was clear Marc was super smart and had such wonderful charisma and drive.  After the devastation of my time at HP, this was so refreshing and fun.  We planned to talk more and see if I could help them as a part time advisor.  That meeting happens in October, 2002 and I'll pick that up in Part 2.