March 2009

Just saw this story on Slashdot which made me think of this post from Marc Fleury.

The Slashdot story questions what/where is the official MySQL tree:

“…With Monty Widenius having left Sun and forked off MySQL for MariaDB, and Brian Aker running the Drizzle fork inside of Sun, where is the official MySQL tree? …. Smugmug’s Don MacAskhill, who is the keynote at the upcoming MySQL Conference, has commented that he is now using the Percona version of MySQL, and is no longer relying on Sun’s.”

In researching a future post I came across this quote from Sun’s Jonathan Schwartz via Marc Fleury’s blog:

“A: First, the personalities involved. Let’s just say that integrating Marten Mickos into a company might be easier than assimilating a few of the JBoss personalities. Marten is a joy to work with and will make this integration work.”

Thinking about JBoss and MySQL today, I can’t help buy chuckle at the quote.  Both JBoss and MySQL have seen their executives leave the acquiring company.  However, for all the “JBoss personalities” out there, not one of them has forked the JBoss code in an attempt to compete against Red Hat.  Not yet at least (you never know what Roy will do ;-).

I truly wonder why.

Maybe JBoss developers are contractually prevented from doing so or have financial incentives not to fork JBoss?  But if so, why wasn’t the MySQL team put under similar legal restrictions and/or incentive plans?

Maybe Red Hat’s history with open source provided for an easier transition than Sun did?  I don’t buy this considering the initial challenges at Red Hat.  Every acquisition has its road bumps, Sun’s challenges with MySQL are no different.

Maybe there were plenty of open source application server choices (Tomcat, Geronimo, Jetty, GlassFish) out there, so the likelihood of a fork gaining traction is too minimal to bother with the effort?  I’m not convinced by this argument; EnterpriseDB/PostgreSQL, Ingres and JavaDB provide enough choice, as do Cloud databases that seem to pop up weekly.

Maybe, as Schwartz foretold, it all came down to the people and the “personalities”, on both sides of the deal

What do you think?

Let me start by saying that Red Hat’s 4Q and full year core business results were solid.  Nicely done.

Full year revenue was up 25% to $653M.  Expectedly, 4Q revenue growth was lower than full year growth, coming in at 18% to $166M.

Also expectedly, Red Hat reported a 27% decline in Net Income (NI aka profit) from $22M in 4Q07 to $16M in 4Q08.  For the full year, NI increased 3% to $78M, well short of the 18% full year increase in revenue.  Fear not, the tepid increase in profit does not point issues with Red Hat’s core business; the sale, service and support of software.

I’d previously covered the fact that nearly half (48%) of Red Hat’s overall profit came from *outside* its core business.  Specifically, the profit came from interest on cash/bonds, from the sale of investments including equity holdings, and gains from currency transactions. Amongst all of these items, interest income appeared to be the largest driver of Red Hat’s non-core-business income.

With interest rates declining to generational lows, interest income would take a hit.  As a result, all things equal, Red Hat’s NI would decline.  Red Hat said as much in its 2008 annual report (pg 48):

“..however interest rate yields on our investments are expected to decline, resulting in reduced interest income”

The profit decline is largely attributable to other income declining from $18M in fiscal 2008 to $5M in fiscal 2007.  This represents a 72% decline in profit from outside of Red Hat’s core-business.  Of this $13M YTY decline, $4.7M was attributable to a one-time gain in the sale of Red Hat’s equity position in a 3rd party company (which appears to have been MySQL).

Redoing my analysis of Red Hat’s “Other Income” we see that Red Hat is coming closer into line with software industry peers.

However, it’s still surprising that over one-third of Red Hat’s profit comes from activities that are outside of their core business.

Wall St. likes to apply a multiple to the company’s profit (per share) to arrive at price targets for the company.  When such a significant portion of a company’s profit comes from byproducts of the core business, simply applying a multiple can be a poor judge of the company’s core business.  But then again, can Wall St. really judge the quality of any company’s core business?

I am not suggesting Red Hat is doing anything wrong or shady.  My findings could be explained by how Red Hat is required to recognize multi-year subscriptions.  For instance, a $5M deal over 5 years would result in Red Hat having to bank $4M in year one and recognize and use it to pay for related costs over the subsequent 4 years. The resulting interest would be counted in “Other Income”.  However, I could argue that IBM’s business, especially the Global Services division, deals with having to recognize large multi-year deals.  And yet, IBM’s Other Income is in the low single-digits. Still scratching my head…

Seeing this tweet from IBM’s Bob Sutor:

“Handled several questions today at #osbc re advice to younger people involved in open source. Makes me feel a bit senior.”

reminded me of a post from early 2007. I’m convinced that “go work at IBM, Microsoft, Oracle, SAP, Tibco, etc.” is as good an answer for those seeking jobs in and around open source as “go work for Red Hat and the like”.

Matt Asay has some good advice for folks asking him for career advice: “Work for an open source company.”

I’d wholeheartedly agree. I’d like to add “or within the open source group of a traditional software company”. (Yeah, Matt’s advice is snappier).

Here’s why I suggest the addition:

Take a look at this job posting for Director of Product Marketing at SocialText which I found on Roberto Galoppini’s blog. The onlyHelp Wanted requirement which appears to be a skill that you’ll build more easily at an open source pure-play than a traditional software vendor is “Holistic understanding of commercial open source software”. If you’re in a division at IBM, Oracle or Sun that is leveraging the value of open source, you’ll get this experience. It doesn’t hurt to play with open source on evenings or weekends.

Next, the following comment from Elias Torres. Elias is an IBMer (whom I’ve never met, but know of because I use his Firefox extension) who worked on Lotus Ventura. Elias is an IBMer working on the Apache Roller open source project and on the team using Roller within Ventura (a commercial IBM product). His experience balancing the needs of the community with the needs of his employer are going to be a valuable asset for Elias I’m sure.

“This is my first IBM product experience and it has been a very positive one so far, but we can talk about that later. The experience of taking something built by the community and making it a product has been incredibly rewarding but brings with it a great sense of responsibility. We are very serious about Roller and are committed to giving back as much as we can.”

Lastly, I can speak from personal experience with trying to get a product (WAS Community Edition, aka WAS CE) based on open source technology (Apache Geronimo) out the door at IBM. Being the WAS CE product manager was the best job I’ve had at IBM. Even though IBM had a lot of experience with open source at the time, a lot of what we did with WAS CE was new territory. Using an open source business model of paid support around a free product was completely new. To say we had some internal angst from other groups at IBM would be an understatement. But the angst wasn’t because we were trying an open source business model. It was because we were doing something that wasn’t business as usual. We got through these challenges with some help from executives that gorked the value of WAS CE to the WebSphere portfolio. The many meetings to explain what we’re doing and how it fits with a given IBM product’s plans also helped. There was, and remains, plenty of day-to-day work to ensure we did/do the right things in the Geronimo community, for WAS CE customers/partners and for our broader WebSphere & IBM customers/partners. Did we make some tradeoffs to balance all three interests? Definitely. Did we usually pick the business as usual route? Nope. Did knowing the business as usual route help? Absolutely. Was it easy? Nope. Was it fun? Heck ya!

I believe that the open source business model will continue to find a home inside more and more traditional software vendors. As this occurs, folks who have experience with open source software and traditional software are going to be very valuable. Folks that can walk the delicate tightrope between the ideologies of both camps will be even more indispensable.

Yes, I know that sounds like a self serving statement :-)

<Update: 2007-01-17> Zend just got themselves a new CEO from BMC. Read the press release or this short interview Matt Asay did with the new CEO and you may notice he’s got a lot of traditional software experience, but very little open source cred. It’s not a negative. Just another reason why I believe it’s good to build your open source cred within a traditional software company. </update>
Follow me on twitter at: SavioRodrigues

PS: I should state: “The postings on this site are my own and don’t necessarily represent IBM’s positions, strategies or opinions.”

Matt Aslett wrote a short but interesting post about transparency in the software industry.  He writes:

“Transparency has become one of the main buzzwords this year and I believe is critically important for traditional proprietary vendors as they attempt to participate with open source in order to reduce costs and benefit from collaborative development projects.”

I was set to agree with Matt’s assertion that transparency is important in our industry.  But I wouldn’t limit transparency to the open source sandbox, nor would I associate transparency with simply making code available (neither did Matt).

For instance, I believe that established vendors can learn from their open source inspired competitors who make project/product roadmaps publicly available.  I was more than a little surprised when IBM’s team decided to externally host their roadmap, and all development activities, including design discussions and decisions.  Speaking to colleagues on the team, the user feedback has been extremely positive (save for the initial comments that the product was not open source, a conscious decision by IBM).

As a product manager, I would love to get ongoing feedback on new features or product improvement requests.  I know that customers want to provide this information; there simply isn’t a scalable way for customers to do so.  Established vendors get this form of product information from high-touch, non-scalable, interactions with their largest/loudest customers.  Putting a product roadmap in the community, openly asking for feedback and then delivering against the feedback could drive efficiency and happiness in the product planning, delivery and usage cycle.  Open source vendors know this and encourage users to get involved with product planning using low-touch techniques (i.e. a wiki).

Encouraging established vendors to open up aspects of their development process will be a key contribution from OSS to the software industry.

I was thinking about the discussion that Rich Sharples (Red Hat), Shaun Connolly (SpringSource), Larry Cable (Oracle/BEA), Jerry Waldorf (Sun), Adam Gross ( and I had at the SD West Application Server panel a week ago.  The session was billed as healthy debate about the Application Server market and our respective company’s role in the future of the market. I’d conversed with a few of these guys in the past and was looking forward to some level of one-upsmanship at the expense of the product/company we were each representing.  However, the discussion turned out to be much more civilized, with little discussion about kinks in the amour of our respective products.  Rather, we spoke about growing the ecosystem in which we all compete.  While this sits well with my Canadian sensibility, I wasn’t sure what to make of it.

Then I saw this tweet from Redmonk’s James Governor:

monkchips its great to see respectful, insightful coverage of Microsoft Silverlight news from #MIX09 from the Adobe team – @mdowney, @sjespers etc.

And then a light bulb went off.

It’s very easy to put down competitor X’s product when your statement is not going to be saved for prosperity on the Internet, and when the people working on product X are faceless.  However, in today’s interconnected market, many of us have interacted online with peers at competitors long before in-person meetings.  As this has happened, I sense we, as vendors in the software market, have raised the bar of mutual respect and learning.  This is goodness for those of us who compete.  It’s also goodness for customers because we vendors are spending our time and resources learning from our competitors in order to deliver a better product experience.  This is significantly more rewarding than spending time and resources putting down other products.  Customers choose products on the strength of the value they deliver, not on the lack of value delivered by competing products.

Dr. Phil over and out.

As much as I’d like to write about the rumored IBM acquisition of Sun, it’s probably best to leave that to others, for now.  I have no knowledge of any real or purported talks between the two companies.  I do know that if the deal goes through, I and many, many, many others will be impacted by it.  Hence, it’s best not to write something on the spur of the moment.

Instead, I’ll talk about Sun’s other reason for being in the news today; their Sun Cloud announcement at CommunityOne.  Aside from the catchy product name, how long did the marketing team work on that one ;-), I’m happy to see Sun move in this direction.  Sun is completely bang on with its vision of:

“a world of many clouds, both public and private, that are open and compatible.”

Sun Cloud won’t GA until later in the summer, so for now, we’re making predictions based on a marketing pitch, and arguably, a demo at CommunityOne (which I am not at).  Sun is promising to do much of what Amazon does today.  Sun believes that its Virtual Data Center (VDC) capabilities will differentiate it from Amazon. According to Sun:

“VDC offers developers a single management interface for staging an application running on OpenSolaris, Linux, and Windows. A drag-and-drop method is used for provisioning compute, storage, and networking resources via a Web browser.”

It’s also interesting that Sun chose to use a Creative Commons license for their Cloud APIs.  However, Glyn Moody tweets:

“Sun’s use of CC’d APIs to create an open ecosystem is interesting; doesn’t mean it’ll succeed, of course…”

I have to agree with Glyn.  With the future of Sun up in the air, it’s difficult for me to see a mass of developers or customers seeking Cloud services turning their back on AWS for Sun.  But, by the time the Sun Cloud becomes available, I suspect/hope Sun’s future as an independent company is more certain. Until then…

Follow me on twitter at: SavioRodrigues

Cloudera, an open source startup working to expand the use of Apache Hadoop, made two announcements today.  First, it has secured $5 million in Series A funding today.  Second, the availability of the Cloudera Distribution for Hadoop.

What’s Hadoop? It’s a platform for developing applications that can process vast datasets while scaling to the levels that companies like Google, Facebook and Yahoo require.  Hadoop is an Apache project that:

“implements MapReduce, using the Hadoop Distributed File System (HDFS) (see figure below.) MapReduce divides applications into many small blocks of work. HDFS creates multiple replicas of data blocks for reliability, placing them on compute nodes around the cluster. MapReduce can then process the data where it is located.”

Cloudera sees a market for Hadoop in enterprise situations from analyzing genome and protein data, oil and gas exploration and financial processing.  The Cloudera Distribution for Hadoop is open source and licensed under the Apache Software License 2.0.  Cloudera intends to drive revenue from support and implementation services.  I’ve typically been down on a support or services-based open source business.  However, in the case of Cloudera, this model makes sense, at least for now.  The number of people who can implement a highly scalable application that processes petabytes of independent data relationship using the MapReduce programming model who don’t work for Google, Yahoo, Facebook and the like can probably be counted on two hands.  There is a degree of education and hand holding that Cloudera needs to do while enterprise developers explore writing this style of applications.

Take a look at the investors and it’s easy to predict that Mike Olson and team won’t be independent for long:

In addition to Accel Partners, investors in Cloudera include Mike Abbott (senior vice president, Palm), David desJardins (early Google employee), Caterina Fake (co-founder, Flickr), David Gerster (entrepreneur), Youssri Helmy (entrepreneur), Dr. Qi Lu (president of the Online Services Group, Microsoft; former executive vice president, Yahoo!), Marten Mickos (former CEO, MySQL), In Sik Rhee (former chief tactician, Opsware; founder, Loudcloud), Jeff Weiner (president, LinkedIn; former senior vice president, Yahoo!), Dick Williams (CEO, Illustra; former CEO, Wily Technology), Gideon Yu (Facebook CFO; former senior vice president, Yahoo!; CFO, YouTube).”

All the best to the Cloudera team.

Next Page »