Jon Mountjoy's Blog

Jon Mountjoy's Homepage

Jon Mountjoy is the editor-in-chief of Dev2Dev and Arch2Arch. Besides managing Dev2Dev and Arch2Arch, he enjoys working with XML, J2EE, Spring and Ruby. He is co-author of WebLogic: The Definitive Guide.

He maintains three blogs on Dev2Dev: the Dev2Dev Editor's Blog about Dev2Dev activities, the Arch2Arch Editor's Blog, and a more personal blog too.

Develop@BEAWorld Barcelona Day Summary

Develop@BEAWorld got off to a great start today with a keynote delivered by Rod Johnson. Rod covered a lot of ground. I enjoyed hearing about the future Java EE 6 and the old stovepipe J2EE architecture. In traditional J2EE (and Java EE 5)—the application server is this huge monolithic piece of infrastructure providing everything under the sun). As Rod puts it, the application server itself offers a set of services. Java EE 6 has features around extensibility and a concept of profiles, and we can now look at an application server as this collection of services. [ Perhaps we can then pick and choose the implementation. I imagine this will change the face of competition among Java EE 6 application server vendors too. ] Rod writes about this in his blog post: Java EE 6 Gets it Right.

Rod also spoke about how the old J2EE was not terribly object oriented because of the "excessive focus on distributed objects." [ Too true, and part of the reason why I love JPA and using Spring. They put the OO back in my apps... ] Rod illustrated this renaissance of true domain modeling with better O/R modeling and better understanding by pointing to two books: the old EJB Design Patterns and the new Domain-Driven Design: Tackling Complexity in the Heart of Software [ Great book ].

Rod also covered OSGi, the various integration points between Spring Framework and WebLogic Server (and Event Server) and more.

I then listened to a talk on Open JPA and Spring, by Patrick Linskey. Patrick introduced JPA and OpenJPA (interestingly, a reasonably number of the audience said they're using Java EE 5), and how to use them from within Spring (for example, by using the wonderfully named LocalContainerEntityManagerFactoryBean. Lots of code goodness.

Patrick also mentioned that they've just working on the next JPA spec release, so if you want to influence it, now is the time.

The next talk I attended was "Ajax Basics and Tools" by Gary Horen. (Unfortunately I missed Eddie's top rated "Basic Technologies and Approaches for Introducing Web 2.0 to the Enterprise"). After introducing Ajax, Gary then spoke about the different programming models: client-centric (application programmers write code that runs in the browser) and server-centric (application programmers write code that runs on the server (which may generate code to run on the client)). I was reminded of Gary's excellent article Exploring Ajax Runtime Offerings, which explores all of this in greater depth.

Gary demoed apps running Dojo and ICEfaces, and I saw (for the first time) Aptana running in his Workshop, allowing JavaScript debugging. Cool. Code completion too.

After running around looking for caffeine, I then attended "Lightweight WebLogic Server" by Jim Gish. Jim pointed out (and demoed) some possible future directions for WebLogic Server. Boy, some of this stuff is neat:

• Installation/download/startup/administration are all a lot faster in the next release. Net installers possibly coming, as well as installers that only install what you need (for example, just the core WLS not the WSRP Producer Support, or Beehive libraries etc.) This extends to runtime too, with optional startup of services to decrease memory footprint and startup times.
• Console is a lot faster and lighter, with a great new look and feel
• Advanced class reshaping. So, for example, you will be able to add/remove static methods and instance methods, adding/removing/changing static, abstract and instance method bodies, adding/removing fields/methods/constructors, etc. etc. in EJBs, servlets, etc. etc. Now this is cool. I wonder if it will get us to a point where developing in Java feels a bit like developing in Ruby (just hit refresh).

All in all, a great day of talks!

Genesis

Alfred Chuang just delivered his keynote, titled "SOA and Beyond: Technology Convergence and Dynamic Applications." The result of users using and creating Mashups means that information is being used differently (and unpredictably, presumably) argues Alfred.

He then argued that business agility is key; businesses don't have time to wait for a year to get a business process or application implemented. Indeed, business processes shouldn't be coded; not agile enough. What we need is to take the elements of mashups and apply them to the business. Alfred calls these "Flexible, dynamic, real time applications." He was careful to point out that they aren't what we traditionally called mashups because they embody business process.

Alfred then announced Project Genesis, an infrastructure platform (extended AquaLogic line) to enable knowledge workers/IT to build apps that fit the way they work. To dynamically construct applications. "Pricing built in", "SaaS", "design for hosted and traditional applications"

We then had a demo of what they had in mind: someone using Workspace 360 (which is integral to Genesis) to look at business from exec level (nice graphs), then click through to business processes, then modify the process by looking up another component to insert (from a repository), then drilling down into the process implementation (looked like WLI to modify it, again using repository), and a promise to drill down even further to code.

In other words, the context can be shared; what the exec wants can be communicated to those doing the business processes, which in turn can be communicated to the architects etc. The shared repository seems to be pretty important here, and the ease with which these different perspectives can share information.

I think I can see where Alfred is going with this. I mean, this is mashups of sorts. It is combining services (although here at different layers). And business processes are key. Mashups for the business I guess. Of course, all the details area little fuzzy, but let's see what turns up!

Live from the showroom floor

Patrick Linskey and yours truly enjoying an espresso earlier

Bridging the gap between people, processes and information

At Participate@BEAWorld Jay Simons gave a great keynote about enterprise and IT. He argues that the inability to launch new business tactics quickly is related to IT integrations that are too rigid, complex and costly. Likewise, the lack of visibility into how the business is run is because of the disconnect between business processes and IT. While these may go without saying, Jay also argues that the business need of knowledge workers seeing greater effectiveness is not being met by current disjointed IT collaboration systems. This last point looks back at what drives businesses. Initially it was the management of physical capital, then later it was information, and now it's interactions.

This is pretty interesting—it smacks of social computing and "enterprise 2.0" - something I blogged about over at Arch2Arch. What's becoming important here are interactions between people and interactions between people and processes. I see this as web 2.0 in a different context—harnessing collective intelligence in the enterprise context.

So what's the future for the enterprise (the "liquid enterprise" as Jay defines it)? Well: Build for change, composite applications, loosely-coupled, business driven (as opposed to application function driven), collaboration ingrained, and heterogeneous horizontal integration. I love the "collaboration ingrained" bit.

In other words, what matters is bridging the gap between people, processes and information. What can deliver on this, Jay argues, is the union of SOA, BPM and enterprise social computing.

enterprise 2.0, SOA, BPM

Keynote BEAWorld Barcelona - NOT

OpenJPA no longer requires bytecode processing (future Kodo too probably)

"Historically, OpenJPA required that you either run a post-compilation tool or run your application with a javaagent. The latest build of OpenJPA removes this restriction by providing various levels of support for unenhanced classes."

Spring Batch

I was burning a backlog of blogs when I came across Spring Batch, which nobody seems to have covered here on Dev2Dev. Batch processing is something a lot of us do, so from the docs: "Spring Batch provides a technical framework and programming model to support long-running processes that perform a given set of tasks repetitively. A typical batch program generally reads a large number of records from a database, file, or queue, processes the data in some fashion, and then writes back data in a modified form. Spring Batch automates this basic batch iteration, providing the capability to process similar transactions as a set, typically in an offline environment without any user interaction. "

Links: Spring Batch Home, Dave's blog post introducing it.

The Building Business Mashups with Java session at JavaOne

I went to (some portion of) an interesting talk on Building Business Mashups with Java given by Adam Gross and Peter Dapkus from Salesforce. Adam spoke about the web services API on Salesforce (they're doing 50M API calls a day) and provided some interesting thoughts on web service creation.

In particular, he believes that if you do SOAP well the potential payoff is greater for end users. Presumably here referring to REST. He pointed out that good web services API design is critical - and due the same consideration as good design anywhere else. Something he recommends is a good distinction between verbs(query) and nouns (contact) - something proselytized by the REST folk too.

Other interesting observations:

• Their platform is multi-tenant. In other words, they have multiple users using the same infrastructure (platform and storage)
• Every customer has their own "schema." Your data model isn't the same as mine. Because the platform generates web services for every customer's unique schema (of objects), this means that every customer has their own WSDL.

These must create some exciting back-end implementation challenges.

Some questions that crossed my mind are: Are they going to start using some of the other WS-* standards like those surrounding trust, transactions and conversation?

Unfortunately at this point the petty security guards ushered me out of the room because I was a fire hazard. It's not clear why. Am I more likely to spontaneously combust? Am I likely to be a hazard to people fleeing the room - those same people currently trapped in the rows of tightly interlocked chairs? Sigh.

Technorati Tags: salesforce, javaone

Dapper Guy at Intel Keynote

I went to the Intel keynote here at JavaOne 2007 - given by Renee James (GM, Software and Solutions Group). It must be a tough gig - motivating Intel to software developers that have purposefully abstracted away the hardware. Having said that, I do have my eye on a Intel Core 2 Duo MacBook (hmm, say santa rosa) - so there is appeal.

After talking a little about chips, Renee spoke about Intel contributions to open source code (including Linux/Virtualization) with over a million lines of code donated to OSS around Java. That's a lot, and I don't know where it is.

Renee also mentioned that they worked with improving the performance of Zimbra, which is Java based. Is Zimbra a pure Java client/server? If so, I don't understand the connection with performance.

A guest, Andy McClure from Visual Numerics, was then brought on stage. These guys do a lot of work with finance service firms. Interestingly their demo running on some 16 core Fujitsu system, used a BEA product.

The next guest to be introduced was Guy Churchward (VP of Java Runtime and Virtualization @ BEA) (dressed rather well) who spoke a little about Java in the data center. During the talk, Guy (and Arvind Jain (soon to be blogger here)) demonstrated "BEA Liquid Operations Control" - something that automates the deployment and governance of Java applications. The demo was quite cool. Using the console Arvind took down a machine - and it automatically restarted somewhere deploying the entire software stack.

Interestingly, this used something called "WebLogic Server Virtual Edition" that gets deployed to a VMWare ESX Server. Liquid Operations seems to be able to control deployments on virtualized resource pools as well as other kinds of pools (a RedHat cluster, I think). The neat thing is that on the hypervisor, WebLogic Server runs on LiquidVM, which is designed to run directly on the hypervisor (think no operating system.) I wrote about some of this last year.

The talk then came back to Sun and their partnership with Intel - I suspect that this might result in improvements to the Sun JVM on the Intel platform. The keynote ended with a pointer to their developer community, which definitely needs a *lot* more pointers back to JRockit.

My final impression was that essentially JRockit is the bridge between Intel and hardware for Java developers - I guess that's not surprising given it's performance lead. For more good JRockit stuff, check out Henrik's blog.

Technorati Tags: JavaOne, Intel, hypervisor, BEA, JRockit

JavaOne Keynote - short stack with coffee

I'm taking a break from my standard Editor's Blog to bring you some coverage from JavaOne. Sitting on the steps here at Moscone, I'm listening to the first keynote. I got in a little late - having a little trouble getting through the "short stack" of pancakes this morning. These initial keynotes tend not to provide too much value as they're often at a pretty high level. So here's a high-level summary.

After some discussion about how communication is an innate feature of humanity, and how JavaME is gonna rule because of the billions of phones running Java, I heard two enterprisey things. a) Ericsson is open sourcing their IMS platform (or something like that) and b) Real Time Java is now ready (what?!)

Unfortunately, details are sparse and Ericsson's website doesn't say much, but I think it's something along the lines of Ericsson open sourcing their IMS implementation and integrating it with Glassfish. Of course, we have on Dev2Dev material on IMS (Enabling the IP Multimedia Sub-system (IMS) with Java Technology and An IMS Application Example Based on SIP Servlets and VoiceXML). I wonder what this means to BEA and in particular, the WebLogic Communications Platform.

The other item announced, Real Time Java, a rather different approach to doing real time than WebLogic Real Time. The way Rich announced it though, it appears to also target some of the same audience such as telecoms and the finance industry.

Also in the news was that Java is now, in terms of open sourcing, done. In addition, JavaFX was announced, a new family of products built on Java (SE I think). They gave a demo of a new scripting language, JavaFX Script, for writing rich Internet applications. It's unclear how this is going to do against Adobe Flex for example. Is this a competitor, an alternative, or is meant for something completely different? Is it going to resurrect Java as an applet?

Sun also announced JavaFX Mobile - a complete desktop-scale operating environment based on JavaSE - for mobile phones (I think) - the background image was very iPhone-esque ;-)

Technorati Tags: JavaFX, Flex, IMS, Ericsson

JavaPolis Day 3: The pain of ironing

I had a good, albeit exhausting, day at JavaPolis on Wednesday. Things started out badly when my hotel couldn't find an iron. A few hours later, they found it without the ironing board. Oh, and it was only marginally larger than my iPod. Seriously. After a frustrating 20 minutes of use, I gave up. Later Martin was kind enough to offer his services to iron my shirts. Wow..

Here's a quick run down of some of the talks that I attended:

• Java Performance Myths by Brian Goetz. One of Brian's points was that something that may well be a performance problem today, probably won't tomorrow. Many of today's performance myths (for example pooling, final methods etc.) are of these type. They were once true. The lesson is that we should rather concentrate on solving the business problems than performance ones, especially at the language/VM layer. It was an entertaining talk, and Brian is an engaging and enthusiastic speaker.
• Failure Happens: Deal with it by Bill Venners. Bill pointed out that there's a human at the top of the error exception handling stack. One good point he made was that sometimes people have too much exception handling. You can't always handle everything, and that's not always a sensible way to fail. Instead, you should sway. His metaphor was of a rigid building in an earthquake. It would fall over, whereas one that can sway, will survive. I enjoyed the talk, but hoped for more.
• The keynotes: Oracle and Sun gave the keynotes. Oracle's was really a product pitch, while Sun's was a bunch of folk showing off Java in small toy devices and extremely weak on content.
• Java EE Enhancements for Real World deployments by Nagesh Susarla. I enjoyed Nagesh's talk, where he spoke about side-by-side deployment, deployment plans (awesome), library management and some interesting things coming up in WebLogic Server 10, including a way to create WLST by recording console actions, and cluster aware EJB timers.

Oh, and during one of the less interesting talks I created my own Flex application.

Technorati Tags: javapolis

POJO Mojo and Sexy Flex

I was minding my own business eating a lunch baguette when I started hearing taunts from Alef Arendsen, ribbing me about my choice of hotel. After a couple of laughs, we started talking about migration and how many folk are still using EJB 2.0 or older technologies. The key, said Alef, is that they have mixed their business logic with their environment. This makes migration very expensive. Insightful fellow that. This is exactly the problem, and perhaps it's even true if you're using (the much improved) EJB 3, where you will bind your business logic with the environment (in this case, the EJB framework). Are you willing to spend a few million euros in a few years migrating to the new EJB 6 framework? Here's my motto: There's no mojo if there's no POJO.

Earlier today I attended the session by Chris Richardson on an Introduction to POJO development, for this very same reason. POJOs, plain old Java objects for the uninitiated, are surely the way to go. With good POJO programming you can write your business logic independent of any framework. Of course, you still need some framework to bind your POJOs together, and to add all the services that you want (transactions, persistence, security, remoting, etc.). But I guess the key here is the POJOs aren't "infected" with these concerns. I can use Spring to provide the services, and later migrate to some other framework if I wish, but my business logic won't have to be touched. As Chris puts it, EJBs lack a separation of concerns, they force you to solve several problems at once: business logic, design, database scheme, security, etc. In contrast, POJO development in Spring for example, can encourage separation of concerns. Perhaps you'll save yourself millions...

BTW, we'll be promoting a chapter from Chris's book here on Dev2Dev soon.

This afternoon I attended Building Sexy Web apps with Flex. That's quite a title to live up to. The speaker, James Ward, promised a look at some BEA portal integration, so I'm reciprocating by attending his talk. And to figure out why I should go back to using an IDE to create clients.

They started with the premise that there is a need for that 'next' form of UI. In the beginning there was the main frame, and then fat clients, and then web applications, and then ....Flex 2? Okay, that may be a little of an exaggeration, but aren't we perhaps starting to push the boundaries of HTML and Ajax just a little. I'm still reeling from those talks I attended yesterday about Dojo - it's like coding in machine code again. Or perhaps it's not that at all. Perhaps people are trying to abuse Ajax and HTML to create an application-centric type of environment, which is where Flex (apparently) comes in.

Flex is certainly sexy. James showed a number of demos (source available) and they all rocked. The way this seems to work is that you write your applications in MXML, an XML language which describes the components and layout. You then link them all together with ActionScript (the scripting language glue). You then throw this through a compiler and produce the product, which is fast. Of course, you need to pay to use their IDE, but the SDK is free and I believe there are some free IDEs being built too.

It's also got some integration with Java, which looks dead simple to use. I'm going to give it a bash and see if I can put something together and blog the result.

BTW. Adobe have also contributed their JavaScript engine (used in their ActionScript engine) to Mozilla (Tamarin Project). It's got a JIT feature too ;-)

At the end of the day, I need to figure out if this is just YAGL (yet another GUI language). Is the pain of using an XML language like that something we should endure? Or perhaps one gets used to it, much like HTML. And what about such languages for constructing fat-clients that folk are developing? Is there going to be a war of the ultimate universal UI language? Is this the main contender? Something to think about while I go and find my frites.

Technorati Tags: javapolis, flex, adobe, pojo

Scripting - where's the PHP?

I ended the first day of JavaPolis by attending Dynamic Languages on the Java Platform. I was motivated to go after seeing all the interesting things you can do with Groovy and Grails in an article a few months back. And the promise of someone saying "duck typing," which cracks me up every time. Charles Nutters and Thomas Enebo of JRuby fame were also present, and gave a great demo of running Rails on JRuby. Unfortunately I didn't get enough information from the introduction about JSR 223, which is what the whole scripting movement on Java is all about.

One speaker pointed to scripting.dev.java.net that lists compatible scripting engines. Boy, was I surprised. The list includes the well known languages such as JRuby and Jython, and many others such as Jawk (!), FreeMarker, Jaskell, Rhino, Jexl, etc. etc.

Speaking of Rhino, someone also pointed me to Project Phobos, which is a web framework with lots of stuff happening in JavaScript (on the server as well). Hmmm.

There was a lot of interest here on scripting languages on the JVM, and rightly so, but one language was glaringly absent. PHP. Why? I don't know - I mean it's only one of the most popular scripting languages around!

Technorati Tags: javapolis, jruby, groovy

Ajax with Dojo at JavaPolis

I thought I'd balance the domain-driven design session here at JavaPolis with something on the complete opposite end of the tier. The talk was Dojo in Action.

A recent article on Dev2Dev, Exploring Ajax Runtime Offerings gave me an inkling of what to expect: a client-centric framework, mostly procedural, bridging everything from individual Ajax components to providing comprehensive framework.

It's frankly a little scary. It even has a little event bus that you can run on the client to provide some nice abstractions (you don't need to know which components are running, or worry about component instantiation order etc.), and does all these weird things with cross-domain scripting and optimization (parallelizing framework loading). I left with the feeling that this is something I'd rather not use—or if I do, get someone else to do it for me. It's pretty low-level hacking, but it also brings up another question: is Ajax the way to go for flexibly, dynamic GUIs?

My JavaPolis registration bag contained a DVD from Adobe for Flex 2. The unfortunate folk over at Adobe neglected to provide the Flex Builder application for my platform (OS X) and only give me a command-line SDK (yeah right), but the intro gives me some insight into what they appear to want, namely to use Flex to build your GUIs.

So, the Ajax folk are advocating enriching the client with Ajax, and the Adobe folk are advocating enriching the client with Flex. Given the complex Ajax I saw today, I can see a role for Flex. But that's proprietary, and for many things simple Ajax is probably sufficient. Like most environments, I think there is space for both approaches. Ajax has the lead though...

Technorati Tags: javapolis, dojo, ajax

Domain-Driven Design: Implications to SOA?

Here I am at the Domain-Driven Design talk by Eric Evans. Its subtitle is "Tackling Complexity in the Heart of Software," which is the key really. As Eric says, "The critical complexity of most software projects is in understanding the business domain itself." Domain-driven design comes into its own when the above statement is true.

It's a well attended talk (about 400 I estimate) and the first half was all about models, domains and how to construct them. I particularly like his definition of a model: "A system of abstractions that describes selected aspects of a domain and can be used to solve problems related to that domain." Note that a model describes selected aspects of a domain, not all of a domain.

Another great quote: "don't ask if a model is good, ask if it's useful." After all, that's the purpose of creating a model. It should be useful, so that you can apply it to solve a problem within the domain. And how do you build a useful model? You find concrete reference scenarios and you have a lot of talks with a domain expert.

Now something that really got me buzzing was something Eric said in passing about reuse. He was talking about context, and how domain-driven design and model development almost always apply to a particular context. A model is used in solving a problem in a particular context. What Eric said is that sometimes these models are wildly successful—the model really works in solving a set of problems. And what happens then? People start using the model out of context, to solve other problems. And what happens then? A lot of ugliness and failures. The model was created for a purpose, and is now being abused because it's being used out of context.

At this point some synapses fired and I thought to myself: isn't this a problem in SOA? You see, even though SOA is service oriented, we also expose a model. Sometimes implicitly and sometimes explicitly. And the model we expose was probably created to solve some particular problem in some domain.

The problem I see is this: SOA encourages reuse, and so the chance of a good model being used out of context is increased dramatically. What are people doing about this? Is anyone working on domain-driven design in an SOA context? Do we need strict governance systems around the reuse of services to prevent models from being abused? Doesn't SOA make domain-driven design more difficult?

Technorati Tags: javapolis, domain driven design, soa

October 2007

Search this blog:

Archives

October 2007
July 2007
May 2007
December 2006
November 2006
October 2006
June 2006
May 2006
April 2006
December 2005
November 2005
October 2005
September 2005
June 2005

Categories

Recent Entries

Develop@BEAWorld Barcelona Day Summary

Genesis

Live from the showroom floor

Articles

Practical Advice on Becoming a Better Blogger
Becoming a better blogger is a snap. This guide provides 11 key steps to improving the way in which you blog. Aug. 8, 2007

Information as a Service
Brad Wright, Product Manager for the BEA AquaLogic Data Services Platform, talks to Arch2Arch about delivering information as a service. The interview covers adopting SOA, integrating data, and accessing data using an information bus. Jun. 12, 2007

Results From the 2006 Dev2Dev Reader Survey
We ran the second Dev2Dev Reader Survey during the last two months of 2006. Here are the results, with commentary. Feb. 7, 2007

All articles by Jon Mountjoy »