I just finished my Data in SOA talk for TechEd Malaysia. The room was very hot and the session was right after lunch – not exactly optimal conditions. I did OK – could have been better. Between time in hotels and on airplanes with nothing to do but code, I’ve written more in the last four days than the previous four months combined. I didn’t spend as much time as I could have reviewing the deck, so I guess I can’t say that I was utterly prepared. But I have delivered this talk many times and spend a significant amount of time thinking about the concepts and discussing them with teammates. BTW, this talk is now available as a whitepaper on Architecture Center.

This morning, Gurpreet delivered his Enterprise Architecture talk. I thought it was pretty good, esp. given that he wrote quite a bit of the deck on the plane to Malaysia. I had only seen him present once before when he was exhausted (hotel screwed up his room) and on pain killers (he fell of the stage and messed up his back). He was much better this time – he’s a great storyteller. He also had a few choice quotes I thought I’d share:

“We could spend the next month in this room talking about enterprise architecture and only just be getting started.”

“Don’t tie your ego to your design.”

“If you don’t do EA, you can’t do SOA.”

He spent most of his time talking general EA topics, with the remainder spent on MSFT incarnations of those topics, such as EASOT. What was funny was that when he showed me his deck, I thought the MSFT stuff play better with the audience, but it turned out the general EA stuff was great. For example, Gurpreet started by talking about the Winchester Mystery House, built by Sarah Winchester (of Winchester Rifles fame). This place is an architectural nightmare with stairs that lead into the ceiling and nearly half of the doors that open onto walls. It’s also over provisioned with 40 bedrooms, 5 kitchens and 17 chimneys. An over-provisioned architectural nightmare? Sounds like a typical enterprise.

He talked about the typical conceptual/logical/physical viewpoint of the enterprise with an interesting twist – the contextual level. This viewpoint is above conceptual in the model. To take his example, if a bank builds an online banking system, we’re all very familiar with the conceptual, logical and physical views of that system. The contextual view might be something like “We’re losing customers due to the fact we don’t have an online banking system.” I need to spend more time thinking about this, and how to map between these views, but I thought the contextual viewpoint was very interesting.

I really wanted to blog what Gurpreet claimed was EA’s biggest fallacy: that it doesn’t change. Because we’ve based the concepts of software architecture on building architecture, there’s this belief that you first design your architecture and then you build things to that architecture – i.e. like a building. For example, when I saw John Zachman present his framework, he was asked how one goes about implementing the framework. His response was something along the lines of: “Build all the layer one models, then build all the layer two models, etc. etc. etc. and then hit compile.” Obviously, that kind of revolutionary waterfall approach to enterprise architecture just won’t work. While buildings evolve and “learn”, they aren’t in a constant state of flux the way enterprises are. This is where Gurpreet made the comment about not tying your ego to a particular architecture – you have to realize it’s going to change.

I’m hoping that Gurpreet’s Architecture Vision & Direction talk will be equally good. Now, I just need to pester him into publishing this material in a whitepaper or blog or something. He’s presenting at Strategic Architect Forum next month, so I expect these talks will continue to improve.

After spending Monday morning exploring Shunjuku, I had meetings and dinner with the local architect evangelists followed Tuesday morning by a meeting with the local evangelism lead. (which was in turn followed by a flight to Kuala Lumpur which is why I’m not posting this until now). In my first year on the job, I focused on the “basics” which included helping improve Architecture Center and handling the TechEd Architecture Track. As part of that focus, I spent little time worrying about globalization issues. Not that they aren’t important, I just didn’t have the bandwidth to handle it. Now that we have worked through the basics, plus it’s now “we” and not just “me”, we can start to have some focus on better globalization support. For example, there’s a Japanese Architecture Center that needs to have the same content as the MSDN site. Machine translation may have come a long way, but there’s still longer to go. Given the massive number of characters in eastern languages, machine translation to English works pretty well – here’s the Japanese Architecture Center in English (translated by Babelfish). However, translating to Japanese (or other eastern languages) is much much more difficult.

Like several other MSFT subsidiaries, Japan has someone responsible for broad reach architecture – i.e. architect with a lower case “a” – so we spent most of our time discussing plans and figuring out how best to leverage each other. I was really impressed how deep these guys were on Software Factories and modeling already – they had a session on Software Factories at the TechEd Japan last week!

Now I’m in Kuala Lumpur for TechEd Malaysia. I’m presenting the Data in SOA talk today and Metropolis tomorrow as part of the Architecture track. I had breakfast with Software Legend Tim Huckaby who I met the first time two years ago at TechEd Malaysia. My teammate Gurpreet just showed me the deck for his Enterprise Architecture talk he’s presenting in about 30 minutes. He’s also doing a presentation tomorrow on Microsoft Architecture Vision & Direction. The EA deck is pretty good and the Vision & Direction talk is something he’s been working on for several months, so I’m looking forward to seeing both of these. We’re working on the next release of The Architecture Strategy Series and I’m thinking these talks should be included.