I was at the Accenture Technology Architecture Workshop last week in London, as an invited member of the panel on "Platform Strategy" along with IBM, Microsoft, BEA, Oracle, Sun and SAP. It was a packed audience of senior management and architects and for once I decided that T-shirt and jeans were not appropriate.

 

For the panel, we were asked to give a brief 5 minute presentation on where we thought THE platform was going. I think we all agreed that there was not such thing as a single platform (another example where one-size does not fit all). There were a few minor differences of opinion on the importance of interoperability and portability. Each of us was asked to say something on SOA and the message was that it's important. At times it felt a bit like a sales pitch from one vendor or another to the audience; "things are great if only you'd use our products and services". I obviously kept Red Hat's offerings in the running, but pointed out that SOA isn't just about software: it's as much about the way in which you think about your systems and architect them. I pointed out that tooling (from all vendors) is still seriously lacking in this regard: it still encourages a closely coupled, distributed object approach for building applications. Not good.

Then we were each asked to respond to the same set of questions:

  • "With the complexity of standards around SOA, and with your company being a primary participant in driving standards, how are you going to make the approach to SOA easier for customers to consume?" Better tooling and education of sales forces/consultants in order to better educate the customer, were pretty much the agreed answers.
  • "How do you view in the market the coexistence of 2 trends, one being Open Source and the other the traditional platform development? How do you take advantage of one or the other? Where is the trend going? What are you doing on Open Source and how (if) do you see this evolving?" Well there were the usual cries of how open source is good for certain things, but closed source was just as viable. It was pretty clear from the audience that they trust open source a lot more these days than they did several years ago. The fact that each of the vendors, even Microsoft, now does some activity in open source, spoke volumes.

 

I got some good vibrations from the panel discussion for our platform strategy as a whole and specifically for the SOA Platform. After the meeting these feelings were confirmed when I spoke with quite a few people from the audience. Everyone seemed to get the whole platform=stability equation. Some even said that it raised the chances of them using Red Hat middleware more now than in the past. I did wonder if I'd get pulled for us not having JBossAS 5.0 GA out yet, but instead concentrating on JBossAS 4.2 and the Enterprise Platform. That question came up twice, but both times it was as a "thank you" related directly to the stability aspect. One person even likened it to XP/2000 versus Vista: "We know we need to get there eventually, but at the moment the 4.x series gives us 95% of what we need now. All we were missing was stability and a solid base on which to deploy for the next 5 years."

 

Over all it was a great conference to attend!

The software industry goes through cycles, in just the same way as other industries. Back in the early 1980's we were working on defining distributed systems, with work on RPC, stub generators, naming services, transactions, security etc. A lot of work was done by vendors and academia, and interoperability or portability were not considered important.

 

Then in the 1990's we started to look at interoperability through CORBA (portability was still mostly a nice-to-have until the end of the century). When Java and J2EE came on the scene portability took centre stage and interoperability (between heterogeneous implementations) dropped away, unless you wanted to embed a CORBA ORB in your application server. Now we're into the Web Services age and interoperability is king and portability is nothing to do with Web Services specifications or standards bodies: that's the domain of implementation and other organizations such as JCP. It can often be a strange and inefficient approach: at least with CORBA you (eventually) got interoperability between different implementations and languages along with portability in the same package. With Web Services, you get interoperability out-of-the-box as soon as you implement (assuming you conform to the specifications and understand what they say, or don't say); but portability may not come for months or years!

 

JBoss/Red Hat is heavily into portability and interoperability, as well as standards compliance. Over the past years we've been looking out for any opportunities to help out customers in these areas, including joining the Interoperability Vendor Alliance, participating in a range of Web Services events, and working on standards committees such as WS-BPEL, WS-TX, WS-Addressing and WS-Policy. Standards compliance is critically important for us and our customers: we want to avoid a return to the bad days of vendor lock-in. We take interoperability as a must when developing our products, rather than as a second thought, by which time it's often difficult to retro-fit for existing customers. Portability is a given because of our strong support for JEE.

 

In that light, we've recently been taking part in a number of important interoperability related events. The Burton Group are hosting the OASIS XACML interoperability workshop and we will be there to participate along with IBM, BEA, Oracle, CA, Securent, Jericho Systems and SymLabs. I'm sure this will be a good event and very successful. Then the W3C has been holding a workshop on transparent eGovernment. Our presentation on eGovernment Portals, is very relevant to our overall goal of interoperability and portability for our customers. Expect more from us around security, federation and transactions over the coming months, as well as interoperability across the board.

Filter Blog

By date:
By tag: