About Me

My Photo
Peter Yared is the CTO/CIO of CBS Interactive, a top ten Internet destination, and was previously the founder and CEO of four enterprise infrastructure companies that were acquired by Sun, VMware, Webtrends and TigerLogic. Peter's software has powered brands from Fidelity to Home Depot to Lady Gaga. At Sun, Peter was the CTO of the Application Server Division and the CTO of the Liberty federated identity consortium. Peter is the inventor of several patents on core Internet infrastructure including federated single sign on and dynamic data requests. Peter began programming games and utilities at age 10, and started his career developing systems for government agencies. Peter regularly writes about technology trends and has written for CNET, the Wall Street Journal, BusinessWeek, AdWeek, VentureBeat and TechCrunch.

Many thanks to Bob Pulgino, Dave Prue, Steve Zocchi and Jean-Louis Gassée for mentoring me over the years.

Monday, July 27, 2009

The Commercial Open Source Failure


This post was also published in Rachael King's BusinessWeek column.

So what happened to open source as a business? There was a wave of “commercial open source” companies that were going to change the world, including SugarCRM, Alfresco, Jasper, Pentaho, and ActiveGrid, a company I started in 2003 to bring open source software into businesses. Each of these companies were going to lower costs with the open source business model and displace existing vendors in categories such as CRM, Document Management, Reporting, and Business Intelligence.

It’s been over six years, and no commercial open source companies other than Red Hat, MySQL, and JBoss have had liquidity events. So what happened? Oracle and IBM, which derive the vast majority of their software revenue from proprietary software, have an increasing share of the software market. And there’s a bunch of commercial open source companies still trudging along.

1. The only successful open source companies sell commodities. Linux, MySQL, and JBoss are the only open source companies with significant liquidity vents, and their success is more indicative that operating systems, databases, and application servers are commodities than that open source is a successful business model. No one can succinctly tell you the difference between Linux and IBM’s AIX, or JBOSS and Oracle WebLogic, so why not buy the cheaper one!

2. It costs as much to deliver open source software as proprietary software. Many open source companies exhibit at tradeshows, have salespeople, systems engineers, customer services departments, and on top of that employ the vast majority of developers that are working on their open source project. So their costs are comparable to proprietary software companies that offer free trial versions for lead generation. It’s great that some random guy in Lithuaniacan fix a bug for the commercial open source company, but the headache of maintaining a community and integrating random code patches is just as expensive as fixing reported bugs with your own people. So commercial open source companies have the same cost structure as the enterprise software companies that preceeded them.

3. Selling software is miserable. Selling software to a large business means approal from architecture committees, security audits, achieving approved vendor status, business unit sponsorship, executive sponsorship, etc. Enterprises see IT as a cost center and IT does whatever it can to not have to deal with yet another software vendor, especially a small startup that is likely to go out of business or get bought and then deal with software that’s no longer supported, making the IT buyer look bad.

4. Customers are switching to SaaS intstead of more software. Let’s say you run a division of a company and you need customer relationship management (CRM) or business intelligence. What are you going to do? Call your IT department so they can find software, purchase it, customize it, deploy it, and then roll out to your users, all while charging your through the nose? Or call a service provider like Salesforce.com or GoodData and be up and going tomorrow? No brainer.

So when is open source successful? When it’s free and supported by a broad community of developers, not just one company trying to extract revenue. At my own company Transpond, where we help companies reach their audiences and customers on social networks and mobile platforms, we use a ton of open source software including Apache, PHP, Tomcat, MySQL, JQuery, Debian Linux, Quartz, Eclipse, Maven, and more, and we haven’t paid anybody anything other than contributing code to some of these projects. The point of open source was for people to share the costs of developing, debugging, and deploying common infrastructure. That does not mean that every successful open source project can sustain a commercial company, especially when they are delivering complicated applications rather than simple plumbing.

8 comments:

Commons Guy said...

"It’s been over six years, and no commercial open source companies other than Red Hat, MySQL, and JBoss have had liquidity events."

IMHO, if you measure success solely by "liquidity events", the sad state of affairs is a reflection upon you and your choice of criteria, more so than it is on "commercial open source".

"but the headache of maintaining a community and integrating random code patches is just as expensive as fixing reported bugs with your own people"

And your proof of this assertion is...what, exactly?

Michael said...

"So commercial open source companies have the same cost structure as the enterprise software companies that preceeded them."

Quite false. The talk I usually give about network multipliers is more along the lines as each developer doing Open Source (commercial or no) can be 5-10x as powerful as a lone developer. The cost of such contributions is quite low, and you miss one of the greatest opportunities -- an extremely rich feedback loop (accessible daily, thank you IRC, you are a godsend) and also a really good community of testers that does not achieve that level of endadgement with commercial software.

If you are a small company, a good model that I like is consulting around the application, and custom development that rolls back into the codebase.

-- Michael (http://michaeldehaan.net)

Peter Yared said...

@commonsguy - when discussing companies that have raised large amounts of venture capital and are running cashflow negative, it is completely reasonable to gauge the success of these companies based on the criteria of venture-backed companies.

@commonsguy & @michael - so i have run and advised both proprietary and commercial open source companies, and they both have pretty much the same r&d, qa, and support headcounts. what's up with that?

Michael said...

I can speak for my personal experience. github impact graphs, sloccount numbers, and so forth. Things that I've had direct experience with -- projects like Cobbler, Puppet, and Func all speak to this working. We are stronger than our paid core. The network multiplier is real. While I agree that there are lots of people that don't give back code, if investment is spent in making sure the community around a project is welcome and helped in learning how to contribute, there are tremendous opportunities that cannot be discounted. And failure in a business model is, well, failure in a business model. I think your list is quite short sighted WRT profitability.

Without seeing your hard numbers, though, we're trading conjecture.

Commons Guy said...

"when discussing companies that have raised large amounts of venture capital and are running cashflow negative, it is completely reasonable to gauge the success of these companies based on the criteria of venture-backed companies."

But you're not discussing "companies that have raised large amounts of venture capital and are running cashflow negative".

What you wrote was:

"It’s been over six years, and no commercial open source companies other than Red Hat, MySQL, and JBoss have had liquidity events."

You are equating "commercial open source companies" with firms that have "raised large amounts of venture capital", etc. You have not provided any evidence that those are, indeed, equivalent. It's as if you were ignoring all of the "commercial open source companies" that didn't fit your argument. If you think there is some obvious canonical definition of "commercial open source company" that everyone universally follows, link to it.

"so i have run and advised both proprietary and commercial open source companies, and they both have pretty much the same r&d, qa, and support headcounts. what's up with that?"

Capitalization apparently is not "up with that", but I digress.

Anecdotes are data points, nothing more. Your personal experiences, whatever they may be, prove little.

Had you phrased this as "in my opinion", or "based on my experience", or the like, I wouldn't be having to lecture you on how to write.

If you are going to make claims such as "the headache of maintaining a community and integrating random code patches is just as expensive as fixing reported bugs with your own people" as being fact, you need to back up those claims with evidence -- surveys, articles, scientific studies, and the like. If you want them to be considered as opinions, express them as opinions.

I just hit on two highlights of your post. The entire post is riddled with unsubstantiated claims (e.g., "The only successful open source companies sell commodities"). The claims may be true; they may be false. However, they are unproven, undocumented, and without any sorts of caveats or scope control (e.g., "of the firms I have had personal dealings with").

Your fast-and-loose statements of "fact" reinforces the stereotype of the Silicon Valley executive airhead. Since I rather doubt that is the opinion you want people to have of you, you might consider taking some greater care with your future posts.

Peter Yared said...

@Commonsguy i think you are picking nits because you don't like the substance of what i wrote, not for real reasons. for example:

i listed examples of the type of commercial open source company i am writing about: SugarCRM, Alfresco, Jasper, Pentaho. they are all venture-backed with the goal to become large well before seven years.

ie: "The only successful open source companies sell commodities": i clearly state that i think commercial success equals a large liquidity event, and the three large liquidity events were all for commodities, as i think i can safely claim that operating systems like linux, databases like mysql, and applications servers like jboss are commodities, because there are many vendors selling identical things.

and yes, as it is a blog, i can state an opinion, which is that the venture-backed commercial open source companies i have interface with have a very similar cost structure to the enterprise software companies that preceded them. why is my opinion valid? i've actually run both types of companies!

ps: there is no need to sling insults or make grammatical corrections about comment threads... keep it civil please.

tycho said...

I think you're getting at a key point of open source business: the success stories aren't the WordPress.com's and the RedHats, and the SunMicrosystems (though they've all been successful by some metrics,) but rather it's all of the people like you (and me, in a very different way) who are able to do good work in "technological applications," using open source, rather than having to write software from the ground up. Even if "The XYZ Project," never turns a profit, it can be successful because of all the people who use to to make value in other contexts... That's huge, and it's hard to draw graphs around, but it's *there*.

Cheers,
tycho

Alice said...

By successful, do you we mean widely used and widely known? While there are many successful open source products, a few stand head and shoulders above the rest.

Greetings,
Locksmiths Bournemouth