During breakfast, I asked Driver if the rumor I had heard about the two-day follow-on conference for next year's Application Development Summit being all about open source was true. He said that there would be a two-day summit on open source before then. It will be held in Orlando in December. That was the first pleasant surprise for the day. Gartner may not be the biggest fan of open source, but it is certainly aware of its presence.
Choosing a development platform
Driver said he had been giving his morning presentation, "Building Software Services in .Net, J2EE, and Open Source," for about five years. The room was nearly filled with attendees. When Dale Vecchio came up to the podium after the talk to ask Driver selected questions from the audience, he noted that every year the angst in the audience over which platform they should use now and in the future was very high. This year was no exception.
During the first part of the talk, Driver noted various statistics from Global 2000 firms on what they are doing today and what they are expected to be doing over the next several years. For example, .Net has been around for a few years, but it seems to have hit a wall. The audience was split about 50-50 between the use of Java and .Net.
For those agonizing over the end-of-life of Visual Basic 6, he said flatly, "You have absolutely no choice in the matter." Neither do add-on vendors enjoy a freedom of choice, and Driver predicted that those who do not embrace .Net are dead.
While Visual Basic is not disappearing, the .Net implementation will change considerably over the next three years as Vista and its new APIs is rolled out. Windows developers have had a fairly stable development platform since Windows 95, but all that is changing with Vista.
Driver said Java is a better choice than Windows Visual-whatever for complex enterprise applications. On the other hand, Driver said Windows development tools are easier and faster for small applications.
He noted that there are very few Global 2000 firms that are pure Microsoft or Java shops. Almost everyone uses both. Shops that manage to get 70 percent of their applications on one platform or the other are doing a good job of managing development costs. Driver said that 80 percent on one platform is excellent.
Driver warned Java developers that applications may not be as portable in the future. In the past, with reasonable care, enterprises could count on 85 to 90 percent of their Java work being portable, allowing them the freedom to choose and mix platforms. He expects that to drop to about 60 percent. The risk of vendor lock-in with Java -- not with Sun and the core of Java, but with add-on vendors adding functionality around Java -- is higher today than it has ever been.
Most of the hour was spent discussing the pros and cons of Microsoft versus Java, but Driver also touched on open source development. He said that open source "has been a godsend" because it allows you to add functionality while avoiding vendor lock-in.
He also noted that Eclipse, an open source development framework, is now "the world's number-one Java development tool."
On the open source event horizon, Driver predicts:
- A complete open source implementation of a Java VM within three years.
- PHP will become a mainstream enterprise tool.
- Mono usage will grow in the enterprise.
- More commercial support for the LAMP stack.
He added a word of caution about Mono, however. Microsoft is happy to see Mono and even Java tools today because they protect the company from further charges of monopoly. Driver said it could crush Mono tomorrow with intellectual property warfare if they wanted to do so, but that Microsoft prefers them let them live for now. The killing blow will come from WinFS and the new Vista APIs. He is very pessimistic about Mono being able to maintain its current high degree of compatibility.
During the Q&A at the end, Dale Vecchio read a question from an attendee about the platform would he choose to develop applications on today, and he was urged not to wimp out by saying "it depends." I was somewhat surprised to hear him say that his choice would be a mix of Java and open source, even though he added that the right choice for him could very well be the wrong choice for others.
Driver's afternoon session was called "The Open Source Scenario." He began by describing the dangers he faced in talking about open source. As an example, he recounted the flames Gartner has received in the past on Slashdot and elsewhere, because they had dared to mention of couple of things about Linux that were not positives along with several that were.
It was a popular session, with 200 to 300 of the show's 700 total attendees packed in the ballroom. Driver began by mentioning that this was not Gartner's first brush with open source. The group did its first symposium on the subject in 1999. Then he made a few predictions:
- By 2010, 75 percent of mainstream IT shops will have a formal open source acquisition policy in place.
- By 2008, open source will compete with closed source in every infrastructure market.
- By 2010, mainstream IT shops will consider open source for 80 percent of their infrastructure software needs.
- By 2010, mainstream IT shops will consider open source for 25 percent of their business software needs.
Next he spent a few minutes trying to clear up any confusion among the crowd on what open source actually is. He mentioned the OSI definition, the fact that it is a licensed model, and the notion of a collaborative community. He also tried to dismiss the idea that open source is something created by a "midnight army of hackers," but did admit that the open source community does include a few ideological extremists.
He explained the phenomenal growth of open source software over the past few years as a combination of things: reaction to proprietary licensing terms, the success of grass roots lobbying efforts, the growth of open source applications outside of the data center, and -- as a result of the Microsoft antitrust case -- the growing awareness of the power held over the industry by just a few companies.
Driver moved on to tackle a number of myths about open source, denouncing as completely untrue the ideas that open source means anti-commercial, or that it is inappropriate for the mainstream, or a passing fad, the idea that there is no control over its development, and finally, the myth that you can't get support for open source products.
To further illustrate the major differences between closed and open source, he talked about the "biology of open source." Driver said open source closely resembles natural selection, the survival of the fittest. The most critical difference, he said, is that in open source, nobody has exclusive control of the code. It's for this very reason, he added, that you are more likely to find a better fit for your needs from an open source project than you are with closed source.
Concerns about the use of open source code
Driver then listed several areas of concern that enterprises should investigate carefully prior to jumping on the open source bandwagon: legal risks, asset management, pedigree of the code, inappropriate total cost of ownership (TCO) conclusions, integration/interoperability with current environment, and migration costs.
He discussed the TCO issue briefly, saying that neither the Microsoft-funded research nor some of the open source backed TCO studies should be relied upon. He also noted, for the record, that Gartner refuses to do funded studies. His point being that TCO varies by shop, and the only reliable way to get a grip on the subject is to run your own numbers.
As he did in the morning session, Dale Vecchio appeared with a handful of questions at the end of the hour. They were swamped with questions, and it was obvious the crowd was hungry for more information about open source. Vecchio admitted that only a few years ago he looked on the open source crowd as "software communists," and added that Driver's talks have been changing his mind on the subject.
Driver's talks changed my mind, too. I admit coming to the summit expecting to hear little or nothing about open source, and I also expected that what little I would hear would be negative and/or misinformed. I am pleased to announce that I was wrong on all accounts. Mark Driver's presentations on open source were both fair and insightful.