BEA eWorld 2004 day-by-day
I have to admit that I will always accept any excuse to visit San Franciso, so I jumped at the chance when my boss asked if anyone should attend this years BEA eWorld.
Compared to Austin, San Francisco's weather is down right chilly. I knew I was going to have a good time as soon as I saw folks pulling on sweaters and jackets. Austin's been realtively cool so far this year, but it's still humid and sticky, so I was really looking forward to the change.
The first day of eWorld was pretty slow for me since I did not sign up for the pre-conference workshops. Instead, I spent the day familiarizing myself with the terrain.
At 5:30 pm the exhibition hall opened to give us our 1st glimpse of what we'd really come for. I'm not refering to anything remotely related to software, I'm talking about toys, T-shirts, pens, majic tricks, free beer and lots of food. We were not disappointed.
At 7 pm we were hearded upstairs to the evening's "Tech Talk". This was actually quite a nice gathering. Several tables were set up, each with a specific discussion theme.
I gravitated to the Open Source table to find out more about BEA's "project beehive". Much of what I wanted to learn will have to wait for tomorrow's keynote (like which existing Open Source umbrella will cover beehive), but it was an interesting discussion and I got to meet fellow java.net blogger Cliff Schmidt.
I've communicated with Cliff via forums, email, and telephone, so it was really nice to meet him face-to-face. Cliff tells me that I just missed java.net editor Daniel Steinberg by a few minutes, so hopefully I'll run into Daniel sometime this week.
The BEA developers that I met tonight seemed very exited to be Open Sourcing their work. They very clearly expressed gratitude for Open Source tools they've incorporated in the past, such as Jakarta Struts, and they were smiling from ear-to-ear that they can now give something back.
I haven't got a clue (yet) on the technical merits of beehive, but I can say that it is a great move for BEA. Sun and IBM have both made significant contributions to Open Source Java, and it looks like BEA is joining the club.
That's all for day one of BEA eWorld. Tomorrow "indoctrination" will begin in earnest, and by the end of the week I'll be chanting "SOA" with a dazed look on my face (Resistance is futile, I will be assimilated).
It's official. BEA's "Beehive" has been accepted as an Apache project. The announcement came during the opening session during which BEA CEO Alfred Chuang made a grand entrance crashing through a wall on a levitating platform. Alfred's wardrobe choice of a Steve Jobs inspired black turtleneck under a cream colored wind breaker was truly stunning. The pep-talk was relatively content-free, but we did learn the new BEA buzzword "Liquid Computing" and the mantra that Compatibility is more important then Integration.
My first "real" session was with Fergal McGovern of SteelTrace on "Mapping Business Requirements to SOA". His advice wasn't limited to SOA; If you don't express business requirements in English and cast them as flows, you won't be able to get the requirements verified by your business owners. Activity diagrams and flow charts will scare most business analysts, and they'll "trust you" rather then checking your work. The tie in to WebLogic and SOA was a demonstration that SteelTrace's Catalyst can generate WebLogic .jpd files. Import these files into WebLogic Workshop, and you have a starting point for your project that came directly from the requirement documents. It's not a full featured solution, but it's a nice start.
Lunch was great. It's going to be hard to make it through the week without gaining weight.
Tim O'Reilly spoke after lunch on the paradyme shift of Open Source.
After Tim O'Reilley's talk I went to Ajay Patel's session on Data Services in SOA. Much of this session was focussed on upcoming features in BEA's Liquid Data product. The most notable change is that Liquid Data will finally be getting full CRUD capabilities (currently it's view only).
From Ajay's session I headed over to Richard Feit's presentation on Struts and Java Page Flows. Java Page Flows is built on Struts, and now that JPF is Open Source it will be interesting to track its wider adoption. To paraphrase "Field of Dreams": "They've built it. Will anybody come?" It looks like good work to me, but it will probably require 3rd party IDE support to take off.
My final session for the day was on Dynamic Aspect Oriented Programming. Jonas Boner and Alexandre Vasseur began with examples of weaving aspects on-the-fly into a running J2EE application (using the WebLogic JMX console). They went on to discuss AspectWerkz and the integration of AOP containers into J2EE. Cool stuff.
This was a good day, and I am looking forward to tomorrow.
Wednesday began with an address by Adam Bosworth. Adam is an entertaining speaker, and shared with us his vision of where computing is heading. In a nutshell, Business Process Modeling is the paradigm necessary to manage the complexity of applications that are spread across the Internet. Loose coupling via an intermediary is necessary to manage dynamically changing services. Metadata becomes pervasive to allow on-the-fly changes.
The next piece of the puzzle is to re-enable Power Users. On the desktop, tools such as Excel, Visual Basic and Power Point have long empowered non-programmers. Adam believes the time is at hand to empower Power Users on the Internet.
Adam went on to talk about User Interfaces (a subject dear to my heart). In Adam's view, self evident user interfaces trump rich GUI user interfaces. Unless you use an application every day, the benefits provided by the rich GUI are offset by having to learn and remember how to use the application.
All this lead to Mobile Computing:
mobile computing provides new UI challenges due to the problem of unreliable connectivity (occasionally connected devices). The answer will be intelligent Browser caching and intelligent UI servers. Adam's son Alex demonstrated a mobile application using a prototype browser and server - no mean feat since the West Coast was experiencing an Internet outage. According to Alex, the application was created using Dreamweaver and XScript with no special considerations for mobility. BEA calls this initiative the Alchemy Universal Client Platform.
Deborah Conrad from Intel followed Adam with her vision for mobile computing. WiFi is changing everthing with the phenominal growth of public hot spots. All computing devices will communicate and all communication devices will compute.
After Deborah, Ron Mak from NASA spoke to us about the 10 principals of Martian enterprise development. WLS 8.1 runs the services layers supporting the Mars rovers. Ron treated us to some photos that NASA would prefer we wouldn't share with the general public, including Alien skulls and Martian cats. Truly a unique way to establish credibility.
My first session of the day was Mike Blevins' talk on Performance Best Practices for WebLogic Integration. Mlke covered both design time best practices and demonstrated run-time pool tuning. Workshop hides a lot of details from the user, but you need to understand these details to get the best performance for your application.
Once again, lunch was great. While stuffing myself on stir-fry, I had a nice chat with an investment analyst and a BEA engineer. None of us could figure out how Open Source makes money, but we sure had fun speculating.
The general session after lunch was a roundtable on SOA best practices. Rhonda Hocker from BEA, Mike Scott from BT, Cindy Dowling from Albertsons, and Jim Gamm from Virgin Mobile shared their SOA experiences from the CIO's vantage. a nutshell, overcoming the Corporate Culture is a bigger challenge then the technology.
From the CIO's I transitioned to the Architects. Will Pugh, Perry Branch, Graham Drury, Erik Frieberg, Bob Hensle, Rag Ramananthan and John Allen fielded questions from the audience. I lobbed a loaded question;"Why will Services be reused more then objects were?". The answer they returned was that services won't be reused unless governence is established within IT to make sure that they are reused. Once again, the technology is only part of the solution.
The session that just finished was entitled "Three Components of SOA: Documents, Processes and Controls". This session, hosted by Garret Conaty, has been my favorite so far. Garret shared with us his insights on what really works when implementing Business Processes with SOA. Pass XML Documents to and from the Services (Controls) that make up you Processes. Garret's excited about this stuff, and it was fun to watch him.
Many in the audience were upset that WebLogic Workshop does not use UML Activity Diagrams to represent processes.
Standardizing Process Diagrams is a good idea, and fortunately, the Business Process Management Initiative (BPMI) has recently released BPMN 1.0 (Business Process Modeling Notation). Both BEA and IBM are BPMI members, so this notation may gain quick traction: Hopefully WebLogic Workshop will adopt BPMN.
As an aside, Stephen White from IBM has done a very thorough comparison of BPMN and UML Activity Diagrams. Hopefully Stephens work will lead to an eventual convergence of BPMN and the UML Activity Diagram.
Now I am off to the session on Beehive. I'm filing this now because my PDA's battery is perilously close to dead, and the search for a free beer will probably be my next priority. All in all another good day.
The last day of eWorld started late. Most attendees went to the baseball game last night (I called it an early evening), so the 1st session didn't start until 9:30.
My day began with Mlchael Rowley's BPELJ session (BPEL rhymes wlth people). BPEL is intended for Web Service Orchestration - BPELJ is intended for Web Service Construction. BPELJ is JSR207 (Process Definition for Java).
I've blogged on BPELJ before, and will no doubt do so again. Michael did a good job of explaining the advantages, but I've got to think about this a bit longer. Do the advantages outweigh the potential for misuse? Workshop will clearly identify what is straight BPEL and what is BPELJ, so perhaps this will be manageable.
I headed from the BPELJ session to listen to Jim Marino and Saurabh Dixit speak on Tuning Service-Oriented Architectures. This talk was based on an actual case study from the automobile industry. Bottom line, the optimal JVM and WebLogic Server settings are very tied to message size. Web Services with large message sizes (over 4 MB) will perform horribly with the default settings.
BEA's original assumption was that most Web Services would have frequent small messages. Experience has taught them that large message sizes are common and must be better addressed in the design of future Web Service containers.
After lunch I went to Michael Cary's Technical Deep Dive on Data Services and XML Query Processing. Michael made a valiant effort to cover XQuery and XDS in glorious detail, and XQuery and XDS are definitely topics that I will want to study in great detail. Note to self: Don't sign up for "Technical Deep Dives" right after lunch (sorry Michael, I zoned out during your talk... Blame it on the big lunch).
From the 21st Century of XQuery, I returned to the mid-20th Century world of COBOL. Acucorp's Sam Tawfik's talk on Enabling SOA through Legacy Modernization covered a subject near and dear to my current employer. Our mission critical apps are all COBOL on the Mainframe. Sam actually managed to give me hope that there may be light at the end of the tunnel (in terms of modernizing our legacy applications).
My final session for this conference was on using WS-Security in WebLogic Workshop. Anurag Pareek and David Remy did a great job covering these topics. WS-Security is probably overkill for most web services today, but David hinted that it will be more compelling in the future.
I'm going to go ahead and close now since I feel that things are quickly winding down around here.
My wife will be joining me tomorrow so that we can spend Memorial Day weekend here in San Francisco. I am looking forward to being a tourist, eating a lot of good food, and generally relaxing. I had a good time at eWorld. I learned a lot, and I got to interact with some very interesting and thoughtful people. As a corporate sponsored event, naturally the information was presented from a biased point of view, but I never heard anyone dismiss or trash BEA's competition. That's a really good thing that I wish other companies would emulate. Good manners generate good kharma, and that benefits everyone.