Wednesday, June 10, 2009

Open Source - No Free Lunch


One of the comments posted on my recent blog titled “Dogmatic Views on Open Source” ( http://tr.im/lUzh ) comes from Peter B…”open source is good, if you don't have any dough, but: there's no free lunch.” There is a lot of meat in Peter’s statement, yet such a statement is no doubt interpreted differently depending on the perspective of the reader. ”Open source is good”. Really? “If you don’t have any dough,” and you select a free option, and fail to achieve your goals because of a poor technology selection, then you have made an expensive mistake. Open source can be good or bad, whether or not you have any dough, and there is no free lunch, but you may end up with a better solution for your needs, and you may end up saving money, or failing. One must first ask, what are we trying to accomplish, and how ready are we to accomplish it?

Three years ago I needed an information only website, and to set up employees email addresses for an IT Governance start-up I was forming, and this was a bootstrap endeavor. I have never written a line of code in my life, nor set up a network more complicated than plugging a router into the cable box in my den, yet via a Google search I selected a hosting company, created an account, entered my credit card number, selected a few options during the registration process, and with only one call for help and 20 minutes assistance from a friend, I had successfully installed the Linux operating system, the Apache Tomcat web server, MySQL01 database and (this is where I needed 20 minutes help) installed the Drupal content management system. All in the cloud, from my home office, with under an hours’ time spent.

It was not free, but it was only about 35 dollars a month, and it enabled me to get a site up in a matter of minutes, and spend 99% of my time figuring out what our content was going to be. In this example, I was paying for web and email hosting, but I was also paying for Dream Host’s ability to offer a reliable technology stack on which to sit Drupal, which just happened to be open source(Linux, Apache, MySQL, PHP, or, the “LAMP” stack). I am not sure what subscriptions Dream Host may have invested in internally for the technologies they employ, open source or otherwise.

So, open source is good if you don’t have any dough. But what about Weather Channel’s CIO Brian Shield, who recently told me he runs an almost exclusively open source shop. He has dough, for him it is just as much about what he is trying to accomplish, and his web site essentially is his business. Employers may hire technologists for their skills in open source, and if a technologist has contributed to open source projects, then she probably commands even more. I would not be surprised if Brian has such talent working for him. I contend that the largest monetization of open source to date is the total of all “premiums” paid to individual developers for their knowledge and contributions either in support or creation of open source code or in helping their employers leverage open source. I contend it will always be that way. I content this is a good thing. I agree with Peter at least on this point, there is no free lunch.

So if my assumptions are true, why all the investment from angels, VC’s, and top software vendors? Open source creators are often highly motivated and highly competent. These contributors want to put out their best, and others in turn want to prove they too have something to offer, even if that is to prove they are smart enough to “break” or find fault in someone else’s creation. This creates a quality control phenomenon that cannot be reproduced inside a corporation…it is a kind of Darwinian mass collaboration fueled by simple economics and a healthy dose of pride of ownership….a great recipe for quality to be created in the most efficient manner, with an assurance of market demand baked in. These developers simply do not in-mass cook up something no one wants but them. Investors encourage [fund] open source development for one reason, to make money, but they attempt to do so in a variety of ways:

A) Drive hardware sales by building goodwill and exposure, while assuring they help drive the standards for which their products are tuned ( Sun created JAVA, IBM helped found Eclipse ).

B) Build infrastructure around which they wrap commercial service subscriptions ( Redhat funds much of the development on Linux Fedora and Jboss.org to create the open source based commercial operating system “RHEL” and the open source based commercial middleware JBoss).

C) Create business applications for which they can offer hosting and/or integrations, customization and support services ( SugarCRM is open source, and they are now a threat to Salesforce.com )

Who else is motivated to encourage open source development? Governments. One example, and you will hear more about this soon: the United States Government and The Open Source Software Institute (http://www.oss-institute.org/). They have opened up a bunch of old government projects because they simply could no longer afford to support them. If you don’t understand why and how this works, reread the above paragraph…I will likely devote a future blog just to this, for now I offer this from their website.

The Open Source Software Institute (OSSI) is a non-profit (501 (c )(6)) organization comprised of corporate, government and academic representatives whose mission is to promote the development and implementation of open-source software solutions within U.S. federal and state government agencies and academic entities.


So how does this work? IBM creates a somewhat captive audience with Eclipse, or at least they often get a shot at hardware sales. If a company is running Linux chances are very high they are paying either Redhat, or in some cases, Novell SuSe ( sometimes even via Microsoft ) for a commercially supported enterprise ready version of Linux. Companies can use the “free” Fedora” Linux, or the “free” openSuSe”, but for the most part, the “cost” of free in this case in too much risk. Most commercial operations acknowledge that the packaging, testing, patching, 24 x 7 X 365 support, standardization and indemnification they get for a subscription is money well spent, and a good bit cheaper than Sun Solaris (now Oracle) or other commercial alternatives. Ok, Solaris has been free for several years, er…but… Sun hardware is not, and I am pretty sure Solarus runs really well on Sun hardware.

As we travel up the stack, the lines get really blurry, and the decisions more divergent on who adopts, who pays, when, and why. JBoss is the most widely used application server on the planet, ahead of both IBM’s Webshere and Oracles’ WebLogic. It is also arguably the best. So why has Redhat stock, who owns the commercialized version of JBoss, not climbed to the stratosphere as companies continue to adopt JBoss for new projects, or even migrate existing applications to JBoss? Because unlike at the operating system level, consumers are much more apt to “go naked”, and use the “free” JBoss.org version instead of the “enterprise ready commercially supported” JBoss.com version. They don’t see as much risk as at the operating system level, and middleware tends to be stable because developers are much less inclined to change their configuration than operations people are to patch an operating system with the latest fix.

In the BI space the decision may be more of a comparison of bells and whistles, with SAP / Business Object, IBM / Cognos, etc. still having more eye candy and then open source contenders. In databases, more likely the aversion to open source data bases has to do with what executives are willing to lay on the line on an open source database…Oracle is still king here, but look for more and more adoption of alternatives databases co-existing along side of Oracle. I can almost hear Larry’s teeth nashing. Why again did he buy Sun, er a, MySql.

To not support open source vendors may be shortsighted, perhaps even tacky, especially if you are transitioning from a closed source solution, or delving into a project that would otherwise not have been funded. This is a case where if you have the dough, you should pony up. By the way, “going naked” is often “offending” the companies own IT Governance, and opening up the organization to legal risks. Besides, who keeps making Redhat products better? Fedora and JBoss.org contributors. Who pays for the contributor’s effort? Largely it is paid employees of a commercial open source company. If you get a “free dinner coupon”, do you still tip the waiter? I hope so.

On the other hand, many companies have looked at JBoss as an alternative to IBM or Oracle and said, “no thank you,” even after a due diligence tells them it is on par with the incumbent technically. They see the elimination of license costs and the reduction in support / subscription cost as interesting, they are even confident that they could make the move, but what holds them back is operational preparedness, or some soft but sometimes very important advantage to lassie fair. Total cost of ownership must be “total”….so what if we do all this and then our developers lose productivity because they now must work in a new and unfamiliar environment. For some even the slight prospect of this in unacceptable. What is a business partner is not on board with the whole open source thing, and you cause a 10 million dollar ripple to save 1 million dollars. Sales reps often cite the motivation for “C” level technology homeostasis as ...”oh, the IBM rep plays golf with the CIO, that is why they won’t go open source”. Maybe, but what information does the CIO gain for that loyalty? Likely it is something more valuable than the savings he would gain for a whole shift to open source. These forces collectively are why all companies fall somewhere on a close source to open source adoption continuum, with most somewhere in the middle. There is a shift, but there will always be a broad spectrum of adoption rates.

Consider the numerous attempts to make money by building companies around the support of open source projects, which have resulted in the creation of companies such as Pentaho and Jaspersoft in BI, Alfresco in content management, and MySQL ( SUN / Oracle ) and Postgres Plus ( EnterpriseDB ) in the database space. Oracle’s motivations may be around control, but the VC’s who investing in these other companies, and the entrepreneurs who founded them did so to make something great and profit greatly from it. It is not a slam dunk, and they too have found, and as Peter reminds us, “there is no free lunch”.


Your comments below are appreciated.