<buzzword-compliant />

test

A pointer from Mike Sax led me to venture capitalist Tim Oren's weblog entry about open source software as a business model. Oren raises the case of MySQL's two-track licensing: a GPL license for open source developers, and a commercial license for those who want to write commercial, non-open source applications with it. The commercial license protects developers against the “viral” nature of the GPL, meaning that anything they do with the code can be kept proprietary. (Novell recently acquired a commercial license of MySQL for its new version of NetWare.)

That's an intriguing approach–one I had been aware of before, but I hadn't really considered the the implications of it. The commercial license includes access to the MySQL JDBC, ODBC and C-based database access drivers from MySQL AB, which are not open source. Developers building pure open source applications can use MySQL freely under the GPL license; anybody who wants to tweak the code of MySQL itself has to buy a commercial license.

Again–the core is GPL; the tools to exploit the core for commercial purposes are not. Open source development is encouraged, while a revenue stream–the real revenue stream, in my mind–is maintained.

Alan Williamson and Simon Phipps are playing a bit of “point-counterpoint” on the profitability of open source. What's interesting here is that Simon, the Sun insider, is the one taking the pro-open source position. [Of course, Simon has been taking that position for quite some time, so it's not really that interesting. ]

I've had a bit of debate with some others over this issue myself. How, one friend asked, can Sun take Java (for example) open source when Jonathan Schwartz is shifting its whole business model toward software?

As Simon says (no pun intended), that question is based on confusion between a development methodology and a business model. Sun owns more than just Java–it owns a stack of software and services built upon Java. The Java language development process is a money-losing effort for Sun–it makes all its money off of the technology that is built on top of that language. So, if Sun were to pull, say, a C# with Java and make the language itself an open standard, while keeping its pieces of the runtime technology proprietary, it would still be able to derive a profit from its products built on Java and improve the cost structure (and the quality) of the maintenance of the underlying programming language.

There are already major chunks of Java technology in the open source domain–Tomcat and JBoss come to mind. But what Sun really needs to do is open up the language itself to the community, while continuing to build a profitable business further up the stack, so to speak.

Over on Sun's Java.net “weblogs” (talk about embracing and extending), Richard Gabriel talks about a recent visit to Sun by Larry Lessig and the “three bears” of intellectual property as they relate to Java…or as Gabriel refers to it, “a fictional…language called *a*a.”

Lessig suggested that the laws of intellectual property would force “*a*a” to one of two extremes–either open source, where its reach could be greatly expanded but the risk of it being “hijacked” or “embraced and extended” [Gabriel's links, not mine ]; or being kept under the control of Sun (whoops, *u*), and thus maximizing its integrity but limiting its potential reach. The third route, or the “just right” that Lessig (and Gabriel, by proxy) endorsed was one where the language was handed over to a “conservancy”, a la The Creative Commons licensing approach–much like what Dave Winer just did with RSS 2.0.

I suspect that Sun will have to take Lessig's advice, or be forced to take Java open-source via some other licensing scheme soon. And I've spoken with enough people who've got some insight to the workings of Sun to be confident that they're certainly already pointed in that direction. Certainly, they've been embracing open source for a while.

The main question is whether they go with a “traditional” open source license or the Creative Commons approach–turning, say, the Java Community Process into , say, an independent, jointly-held (with the other Java contributors) intellectual property holding corporation , something I said they should do nearly 4 years ago when I was at Java Pro. (Unfortunately, that opinion is hidden behind a pay-to-read firewall at DevX.)

Unless they do, Java's not going anywhere. Many of the major contributors to the Java code base are reluctant to cough up more of their hard work to add to Sun's bottom line; for Java to explode into the mobile market and other spaces where Sun has no current clout, it really needs happy (and unfettered) partners to make it happen. Microsoft is already moving in a crushing fashion into the wireless market, making some sort of Java “glasnost” an even more urgent requirement.

But then there's that damn lawsuit…and there's the beating Sun is taking with its sales numbers. So the question is, will Sun jump out of the Java frying pan and into the revenue fire? Or does McNealy understand that he gains more strategically by coughing up Java to a separate licensing company and sharing the wealth?

I've talked a couple of times recently to a company called Netli. If you haven't heard of them, or their NetLightning service…well, you just did. And you're bound to hear more about them, or folks who try to copy what they're doing–building a network of what amounts to performance wormholes for Internet applications.

The concept is pretty straightforward: by building a set of access points across the globe connected by a high-speed network not hobbled by the performance issues of Internet Protocol, they've created “virtual data centers” for their customers that bypass the latency of Internet routes and boost the performance of web applications for distant customers–not just web pages, like content staging services do.

To do this, they've created a high-performance data protocol of their own that web traffic is tunnelled over, and access points that turn the data back into standard Internet traffic at the local ends. Using DNS redirection based on geography, users in, say, Japan get pointed at Netli's point of access in Japan instead of following the Internet's routes across the Pacific. Instead of it taking up to 10 seconds to load a page, it takes less than one.

What's cool about this is that, unlike content staging services like Inktomi, Netli can deliver web pages reliant on real-time data–like ASP applications, corporate portals, and web services. The alternative for most companies would be to have regional datacenters serving up those applications–but the challenges of keeping them all synchronized, even with a private backend WAN, are daunting.

Netli just inked a deal with IIJ to cobrand the service in Japan, and is already delivering the service to North America and Europe. HP is using Netli for its Asian developer portals.

NetLightning addresses one of the problems of the “World-Wide” part of the WWW. Because of its DNS sleight-of-hand, it's transparent to people using it. And Netli is looking at other Internet applications–like, say, virtual private networks–as future services.

For those applications, it remains to be seen how Netli's proprietary pipe compares to RouteScience's optimized routing in terms of performance and bang/buck. I'll be digging some more on this.

Judge satisfied on Microsoft antitrust compliance. CNET Jul 24 2003 12:22PM ET [Moreover - CNET]

Microsoft cuts the entry fee for looking at its server protocol hooks from $100,000 to $50,000. Well, at least we all know which IP ports those protocols use, based on the security problems reported last week. Maybe Microsoft should just issue the protocols as a vulnerability report and get it over with.

The Trackback Development Movable Type weblog reports the trackback capability in Manila. I'm still trying to verify that it works in Radio.

I see Mark Pilgrim has started including a revision history for each post on his weblog. This is an interesting feature, especially if you've got a group weblog or a blog tracking the status of several issues–you can bookmark the permalink of the entry, and watch that specifically, rather than going through the whole reverse-calendar order of the site to find updates. Very cool.

DVD's Meant for Buying but Not for Keeping. The Walt Disney Company's home video division plans to test market a DVD that never needs to be returned because it stops working after a fixed period of time. By Eric A. Taub. [New York Times: NYT HomePage]

Two groups that will hate this–video rental stores and environmentalists. Another form of throwaway culture is all we need–the DVDs should come with a return deposit for recycling, at least.

SCO says it's time for Linux users to pay up. [The Register]

Ya gotta wonder what's going on at SCO Group. The company formerly known as Caldera, erstwhile Linux OS distributor, is clearly attempting to precipitate a crisis in open source land that, sadly, many could have predicted when Linux started to attract money a few years ago–a crisis that Caldera played an active part in until its sudden personality change as holder of the crown jewels of Unix System V.

BSD is looking better every minute–though I suppose SCO is looking for ways to sue anyone using the Mach kernel, too.

Did IBM and others dump intellectual property that didn't completely belong to them into the open source code of Linux? Did they, in fact, give Linus Torvalds a poison pill of multiprocessor code?

For most people, the question, as the linked article in The Register suggests, is moot. If you're only using single or dual processor systems in cluster to run Linux, the alleged infringed code may not be of any consequence. That means that the only people really impacted by SCO's claims may be those trying to run Linux on a mainframe or on IBM or HP mega-servers.

And there ain't a whole lot of those people out there.