Slowly but surely, open source is turning our industry on its ear. As a result of this trend, I've been wondering whether IT managers and solution providers have contingency plans for the very different future that could be in the cards.
Over the last few years, open source--meaning the royalty-free deployment of commercially capable and modifiable source code---has been doing quite well against commercial alternatives. Combine the success of open source software with the current major architectural revolution--service oriented architectures (SOA)--and we may be headed into a world where the only things we pay for (and that solution providers can make money on) are hardware, outsourced services, and patents (via royalties).
SOAs push all development up and away from the operating system and into the middleware layer. Given the increased focus on the two viable technologies for such a middleware layer--Sun's Java and Microsoft's .Net--the operating system itself is becoming increasingly irrelevant as a development platform. Not only that, but the playing field is slowly being leveled across the remaining differentiators such as reliability, performance, manageability, and security, to the point that operating systems could become little more than a commoditized kernel of low-level functionality.
That trend alone does not bode well for solution providers who derive a significant portion of their profits from operating system licensing fees. But once an open-source alternative becomes a technically viable alternative in such a commoditized market (as Linux and Apache have), then all bets are off. Microsoft clearly acknowledged the threat of open source to Windows in the antitrust proceedings against it.
The question now for IT managers and solution providers is whether this open source effect will follow developers right into the middleware layer where Java and .Net are positioned to rule. If technically and legally viable open source alternatives to fee-based Java 2 Enterprise Edition-based (J2EE) or .Net-based middleware foundations become a reality, then the industry will be one step closer to a world where licensed mass market software could be relegated to the past. Already, open source alternatives (JBoss) to fee-based J2EE-based application servers (IBM's WebSphere, BEA's WebLogic, Oracle's 9iAS, etc.) exist, and Sun is giving no indication that it's prepared to enforce the intellectual property rights to Java that would otherwise prevent such developments.
Sun is between a rock and hard place when it comes to its Java patents because of the way open-source projects like Apache (a kissing cousin of Java) are involved with the Java Community Process (JCP), which sets most Java specification standards. On one hand, given the importance of Apache to Java developers and many mission critical Web-based applications that exist today, Sun has little choice but to embrace the idea of open source.
On the other hand, after embracing open source to serve one interest, Sun runs the risk of looking hypocritical if it clamps down on open source to serve another. In Sun's case, the patent cat is out of the bag, and there isn't much chance of putting it back in. Sun executive vice president of software Jonathan Schwartz isn't concerned, though. In a recent interview, Schwartz said that he believes IT managers won't settle for anything less than J2EE servers that are certified to be compliant with the JCP's specifications and that open source implementations like JBOSS can't get such certification. However, Schwartz might want to tell that to companies like General Electric or FCCI that have taken the JBoss plunge.
On the .Net side, the likelihood of an open source implementation isn't nearly as clear. Given the way Microsoft has vociferously enforced its licensing through anti-piracy efforts and technologies, it stands to reason that we may never see an open source implementation of .Net. much the same way we haven't seen one of Windows. That said, Microsoft is dipping its toes into the non-Windows water by turning a very thin slice of the .Net specification into ECMA and ISO standards. Theoretically, the existence of such standards could lead to implementations on any operating system, not just Windows. But, even if that happens, there would still be a huge gap in functionality between such implementations and a full-blown .Net-based application server. It's not like one application server could be substituted for the other the way JBoss could be substituted for WebLogic or WebSphere.
Microsoft has also released a BSD-Unix based application server called Rotor. Like the ECMA and ISO specifications, Rotor represents a minimal subset of the functionality found in a full-blown .Net-based application server and cannot be considered a .Net substitute. Even if, technologically, it could be such a substitute, Microsoft's "shared-source" licensing terms prohibit commercial usage of the technology, relegating it to usage by the academic community for research purposes.
Where Microsoft appears unwilling to embrace production of an open-source alternative to .Net, others are willing to step in. The first company to test Microsoft patents to .Net in the way that JBOSS tested Sun's patents to Java will very likely be Ximian. While Ximian has been vowing to release Mono, a Linux-based, fully .Net-functional open source alternative to Microsoft's Windows-based application server, the industry has been holding its breath to see what Microsoft will do about it.
Microsoft could do one of three things. It could attempt to enforce its intellectual property rights by preventing Ximian from distributing Mono. Or it could demand royalties from Ximian. The third option, and the long shot that I would bet on, is that Microsoft will acquire Ximian. Eventually, Microsoft might recognize that, from a market share perspective , the benefits of a Linux-based version of .Net outweigh the risks it presents to the Windows franchise. After all, the action (the revenue) is where the developers are at, and the industry-wide emphasis on service-oriented architectures has pushed developers to the application server layer.
Encouraged by Linux, Unix, or any other non-Windows operating system support, many Java developers who have so far turned their backs on .Net might reconsider their options should Microsoft pursue operating system agnosticism.
 | ||||
|
|
| ||
|
Reader Resources  Open
SourceZDNet White Papers |
| ||
| | ||||
| | ||||
Citing development work that's left to do on the user interface front, Ximian co-founder and CTO Miguel de Icaza told me that "Mono is about a year away from becoming 100-percent compatible with .Net." De Icaza went on to say, "We'll distribute it as open source, and we're not worried about Microsoft enforcing its intellectual property rights. That's because any relevant patents are in .Net's core and those parts were made royalty-free by virtue of the ECMA standardization process. Even if Microsoft did attempt to stop us--which they haven't so far--they'd have to prove that their claim is valid and that prior art existed. On the issue of prior art, if our implementation were proven to duplicate previous work, we'd find a way to implement the same functionality with a different technique. But, before that, Microsoft might have to prove that Java isn't a form of prior art that invalidates Microsoft's patent claims. The core invention, which is what ECMA standardized on, looks very much to be derived from Java. I think we're in good shape."
When asked about whether it might make sense for Microsoft to become a Ximian stakeholder, de Icaza said that "people would complain, but, to the extent that .Net's growth depends on an OS-agnostic philosophy, it makes sense."
Should open source successfully penetrate the application server layer, the industry's power structure could be very different from what have today. The more that traditional software companies--those that depended on revenues from licenses--start to give away software, the more they'll have to focus on the remaining havens of revenue for significant revenue: services, hardware, and in some cases, patents.
To some extent, the industry appears to be girding for this eventuality. It's hard not to notice how companies like BEA have been giving away software through downloads in hopes of picking up support-based services contracts on the tail end. In a recent interview, BEA CEO Alfred Chuang, said, "I saw a report coming out from one of our financial analysts that shows there are millions of downloads of JBoss. And obviously everybody knows there are many, many millions of downloads of [BEA's] WebLogic to the point that it becomes irrelevant for us to really publish those numbers any more--people believe those. But [JBoss] has only 75 paid customers for support. BEA has 14,000 paid customers for support."
BEA isn't the only example. This free software mode of operation is increasingly more commonplace. Sun, for example, is including more and more software as part of its operating system. Red Hat's business model, where its distribution of Linux is freely downloadable while the company offers a menu of valuable services, assumes that software licenses can't be depended on to keep the company in business.
When it comes to companies that are well positioned to capitalize on a world where services, hardware, and patents drive profits, IBM is in a strong position. As an ardent promoter of open source, IBM gives its customers the option of buying just about any kind of hardware (everything from PCs to Unix systems to mainframes) and doing the rest of the work themselves, or turning over the marriage of the application and hardware infrastructures to Big Blue's cash cow consulting arm--IBM Global Services (IGS).
Proving that businesses are embracing this model, barely a week passes without some consultancy (IGS, EDS, HP, Unisys and others) announcing a multi-billion dollar outsourcing deal. IBM is also rumored to generate more than $1 billion per year from its patent portfolio.
As I have posited before, the annuities from patent portfolios are so attractive, it's not surprising that Microsoft is trying to maintain an intellectual property stake in some key Web services specifications. If a business model that's based on software licensing fees falls out of any favor, the company will need to find other ways to diversify its revenue base. If hardware, services, and patent portfolios are the only safe havens, then Microsoft could be for a bumpy ride since it currently has little to offer in those areas.
In his discussion of the next big software transition, my colleague Dan Farber draws attention to Microsoft's service-based foray into the customer relationship management space. On-line gaming services such as XBox Live and financial services like MSN MoneyCentral are other examples of Microsoft's diversification into services. It's more evidence that services could be the name of the game for both vendors and IT managers in the future.
As I said earlier, hardware is also an interesting haven in this new world. Hardware can't be open sourced. IT managers can avoid buying hardware by outsourcing their applications, but somewhere, someone has to buy the hardware that the software runs on.
The question for IT managers is: What providers are positioned long term to survive in an economy where the margins on hardware can be razor thin? Here again, IBM is well positioned because it can package hardware and services in a way that it can take a loss on the hardware, but more than make up for that shortfall on the services side. HP, with its consulting practice, is also well positioned for such a business model. Hardware pure-plays like Dell and Intel can also tap a similar vein.
Dell has continued to report profits at a time when other hardware companies are suffering. The company has been gradually entering the $350 billion professional services business, and it ranks as Dell's fastest-growing business unit and has the highest profit margins.
No matter what happens to the industry's structure, the one company that's virtually guaranteed to do well is Intel. It doesn't matter whether we go the licensed software route or the services route, or what application server we pick, or what operating systems they run on. Most roads point to Intel. If Java wins, Intel wins. If .Net wins, Intel wins. If Windows wins, Intel wins. If Linux wins, Intel wins.
About the only market shift that could derail Intel would be if AMD trumped Intel on the 64-bit front. Whereas great compatibility exists between their respective 32-bit offerings, there is no such compatibility on the 64-bit front where AMD's Opteron is hoping to give Intel's Itanium a run for its money. If either company emerges victorious, that victory will be preserved by-- you guessed it-- the patent portfolio. To a large extent, the legal existence of Intel clones came as a result of ancient cross-licensing deals that Intel could not undo, and that paved the way for clones to exist. Given that Intel's and AMD's incompatible 64-bit instruction sets represent a fresh start for both companies, ancient cross-licensing deals are not likely to loosen either company's exclusive grip on their own patents.
The other competition Intel has on the hardware front is from the two main purveyors of non-Intel-based operating systems--IBM and Sun. While HP has already committed to an Intel direction for it's major operating systems (HP-UX and OpenVMS), Sun will continue to provide its version of Unix (Solaris) on SPARC hardware. But it will also be providing Solaris and Linux on Intel (or Intel-compatible) hardware. Meanwhile, IBM, which sells a lot of Intel-based hardware, continues to make investments in PowerPC, the microprocessor behind its non-Intel systems. In Intel's favor again, IBM Software Group Senior Vice President Steve Mills recently stated that Linux is the likely successor to IBM's version of Unix (AIX), which currently runs on PowerPC hardware. Linux runs on Intel. AIX does not.
Should mass-market licensed software one day be a thing of the past, there's no time like the present to consider the ramifications on everything from your careers to your IT strategies, to your existing relationships with solution providers. If the industry eventually flips on its ear in this way (admittedly, a big if), it may take a while to play out because a lot of software companies will kick and scream as they get dragged into the new world order. But sooner or later, they'll have to go.
Leave a message in our TalkBack forum or e-mail me at david.berlind@cnet.com.
