02.25.02
Posted in .NET, Application Development, Java at 6:15 pm by Tony Baer
Over the past few weeks, we’ve discussed a lot of the cool new products that have rolled out in run-up to Microsoft’s Visual Studio .NET release. All of them, VS.NET included, have been designed to make complex application development tasks easier than ever before.
Our basic question: Is that such a good idea?
We posed this question a couple weeks back when we reviewed Rational’s new XDE UML modeling tool, which generates complex UML models while you code in VS.NET. For the record, we want to correct ourselves here: Rational wasn’t the first to offer UML support for .NET languages, TogetherSoft released such a product last fall.
Back to our point. The original idea of developing object models in UML language was to build reusable application frameworks before coders got to them. But these tools allow models to be generated as afterthoughts during coding. Will this get developers up to speed on object development or simply give them a weapon to become more dangerous?
Today’s BEA release of WebLogic Workshop brings up similar questions. BEA is trying to solve the problem that the richness of the J2EE framework, which made it the gold standard of web application development, also made it hard to learn. Even today, few developers have managed to tackle Enterprise Java Beans (EJBs), arguably the brass ring of J2EE.
BEA WebLogic Workshop is developed like the old 4GLs, providing an event-based, visual approach to deploying sophisticated services like data caching, system failover, messaging, and transaction pooling. Admittedly, BEA has built many defaults into WebLogic Workshop to steer developers to making the right configuration choices. Will this help client/server developers master the intricacies of J2EE?
Why are we having this conversation? Even with the technology market downturn, skill shortages persist. Visualize the talent pool as a pyramid: on top is an elite cadre of enterprise architects, designers, and analysts who know enterprise business rules, the underlying plumbing of enterprise systems, and OO design. They remain highly paid and hard to find. Below them are throngs of lower-paid developers who code and can be bought by the dozen.
Similar questions arose a decade ago when Visual Basic opened up programming to liberal arts folks, allowing many companies to pry open the riches of their back end systems. Maybe it’s not a stretch to expect that the new tools could help client/server developers grow up, but we’re still keeping a wary eye.
Permalink
02.19.02
Posted in Legacy Systems, OS/Platforms at 6:14 pm by Tony Baer
As we noted a few weeks back, Linux markets are entering adolescence, with platform vendors assuming the initiative from back room geeks. Like UNIX before it, platform vendors are pushing, or being pushed by, popular demand.
Sun is being pushed. Finally biting the bullet last week, Sun announced it would open up SPARC servers to Linux. Conversely, IBM is treating Linux as new opportunity, witness today’s announcement of the new z800 series. A new “mini mainframe” with 10 models, z800 offers two OS options: almost Pure Linux, or a ‘lite’ version of z/OS called z/OS.e. IBM is slashing prices dramatically on the new boxes, starting at under $400k for hardware and three years maintenance.
That’s not as brave as it sounds, because, by omitting support for core CICS and IMS transaction processing applications, the bargain basement z/800s are aimed strictly at what IBM calls “new workloads”: web serving, file and print serving, and enterprise applications, that largely require UNIX, not mainframe professional skills. To us, the obvious market looks like service providers, not data centers.
What a deal, or is it? Although Linux looks like UNIX, running z/800 boxes still requires some classic mainframe administrator skills because they utilize z/VM, an underlying piece of mainframe OS. Then there are software vendor licensing practices to contend with. Remember, IBM has been chopping mainframe prices for years, but ISVs have typically reclaimed those savings with classic MIPS pricing schemes that raised licensing fees with every box upgrade.
Both hurdles are clearly surmountable. On the technical side, IBM’s project eLiza, aimed at making mainframes self-managing and self- healing, could reduce mainframe skills requirements. And on the marketing side, if IBM could persuade mainframe ISVs to adopt the CPU-based licensing schemes that prevail in UNIX/NT, software costs will grow more competitive. IBM is holding up the magic carrot of services to ISVs (this market will certainly need them), but fulfilling that promise will dictate Herculean efforts for keeping IBM Global Services at bay.
That’s a lot of ifs, but then, we said similar things about UNIX a decade ago.
Permalink
02.13.02
Posted in .NET, Application Development, Java at 6:12 pm by Tony Baer
Over the past few weeks, we’ve been observing a flurry of vendor announcement activity spurred by Microsoft’s forthcoming Visual Studio .NET (VS.NET) launch. Well, it finally happened. OK, we weren’t in San Francisco for the event. But it was just a formality, since VS.NET has been available in a highly public beta to over a million developers since last summer. Little mystery there.
But the release of VS.NET provides a real illustration of what a little competition can do, given the challenge of Java.
Make no mistake, Java was a tough act to follow. Forget the client-side failures, Java’s reputation was built on the server. Java’s J2EE server framework accomplished what the Object Management Group’s CORBA never could: deliver a platform-neutral application model, supporting distributed architectures with the scalability that client/server lacked. Yes, parts of J2EE have proven hard to implement, but J2EE applications are scaling as advertised.
Microsoft rose to the challenge with the .NET Framework, which handles most of the housekeeping chores performed by J2EE. Built around XML and web services, .NET substitutes Java’s platform-independence with language-independence. The result? There’s new competition in town, with Java playing catchup in supporting web services.
VS.NET cements .NET’s gains with development tools sporting Microsoft’s famed ease of use, but taken up a notch. For starters, all of Microsoft’s languages are unified under the same front end, and thanks to published APIs, third parties like Compuware and Rational can blend their offerings under the familiar Visual Studio front end. These advances will help project teams with varying skillsets work more closely in unison.
But, just as Java’s write once run anywhere promise proved elusive, chances are, getting all non-Microsoft .NET languages to interoperate in perfect harmony will hit similar speed bumps. Another major challenge: Getting legions of Visual Basic developers up to speed on .NET, a much different creature from what they’re used to.
So now it’s Sun’s turn to flaunt the glories of competition.
Permalink
02.06.02
Posted in .NET, Application Development, Java, SOA & Web Services at 6:10 pm by Tony Baer
The dog and pony shows just keep on coming. Just today, Rational and Sun made announcements–in one case, to capitalize on the upcoming release of Microsoft’s Visual Studio.NET Integrated Development Environment (IDE), and in the other, to overshadow it. Recent days have also seen similar announcements from BEA, Sybase, and emerging players like new Microsoft partner Avinon (more about them next week).
It’s all about staking claim to the as-yet unborn XML web services marketplace. Last week, Sun had a J2EE partner certification party, and this week, followed up by releasing a J2EE “Web Services Pack” that further attempts to preempt Microsoft. Just as Microsoft released public betas of VS.NET, Sun didn’t feel the need to wait for JCP final approval before issuing this pack.
To customers, Sun’s rollout makes little apparent difference since most J2EE middleware and tool vendors already supply their own web services toolsets (the lone exception, BEA, just rolled out Cajun, and promises to announce more at its user conference later this month). The only difference is that each vendor will swap out interim APIs under the hood.
Sybase, in turn, recently unveiled its long-awaited integration of the EAI tools from last spring’s New Era of Networks acquisition. For now, it ties the EAI broker to the J2EE appserver, something that rivals like SeeBeyond already do. Given Sybase’s large spectrum of application development tools, we’re interested in future possibilities that might see the business process modeling features of its IDE and EAI tools converged to bring business analysts and application developers closer together.
That’s indirectly what Rational had in mind with XDE, its follow-up to the company’s established ROSE UML modeling tool. Having dominated mindshare with UML–three of the company’s scientists helped write the language–ROSE has always been awkward to use, and was largely restricted to architects and designers. XDE is not just a facelift of ROSE, it literally rips ROSE’s skin off by functioning as a transparent extension of the IDE itself. As you write code, the UML model automatically generates. While TogetherSoft has had this capability in its UML-based Java development tool for several years, Rational is the first to serve both the Java and .NET camps (XDE is available for IBM’s open source Eclipse/WebSphere Application Developer and Microsoft’s VS.NET).
It’s always great to make things easier, but we wonder if Rational XDE– designed to bring UML to the masses–will either crowd traditional architects out, or bring developers and designers together.
Permalink
02.04.02
Posted in Linux, OS/Platforms at 6:07 pm by Tony Baer
t’s been at least 12 – 18 months since the Linux bubble popped. What remains of Linux (1) increasingly resembles the UNIX marketplace and (2) exists almost solely at the pleasure of the major platform players.
Point one, providers of Linux distributions are making larger chunks of their technology proprietary, just like the UNIX folks. For instance, while Red Hat’s scalability enhancements are still offered under the customary open source General Public License (GPL), they will be very hard for community members to just download. Fortunately, Linux creator Linus Torvalds is still around, because otherwise he would probably be rolling over in his grave.
That places Linux vendors in a conundrum. If big chunks of the technology become off limits, it will be harder for them to improve their products. The Linux guys don’t have the R&D depth of IBM, Microsoft, or Sun.
Point two, platforms are going to drive the Linux business. Sure, the Linux folks want services businesses, but the track records of VA Linux (now VA Software) and Linuxcare offer scant encouragement. Instead, the hardware folks are propping up Red Hat, Caldera, SuSe et al because they don’t want the Linux distribution business.
IBM and HP made the loudest noises. IBM, whose Linux TV commercials are pretty cool, unveiled a mainframe that was pure Linux, sort of. It still requires IBM proprietary technology to partition the box. But, to run what? Web and print serving or email? We’d rather see IBM try running classic MVS images atop Linux. Sure, that might cut marginal software revenues, but IBM makes its money on services anyway.
We liked HP’s initiative to get carrier-grade Linux boxes for telcos, whose current troubles make them receptive to cheaper solutions. Now, here’s where Sun should get creative. Sure, their Solaris boxes dominate telcos, but why not fight fire with fire? They’ve got the SPARC folks, who know big systems, and a small Cobalt subsidiary that makes Linux appliances. Why not bang heads together and roll out some Cobalts on steroids? It might cannibalize some telco SPARC markets, but it would sure keep HP out of Sun’s turf.
Permalink