Steve Craggs picked up on IONA’s recent announcement, in which we reiterated our incremental adoption approach, asking whether this is a good thing. Neal Keneally has already commented that if you know you need orchestration, you can always buy it at the same time as the Artix ESB microkernal, and Steve apparently ended up agreeing with this, but still wondering how the final tally would come out. Would it really be less expensive than buying another ESB the “traditional” way, all at once?
This is a good question, of course, and one that’s very difficult to answer because in the world of enterprise software, sales are typically a long discussion, including price negotiation. It’s not like buying Word or Flight Simulator.
But I would add that SOA is fundamentally different than previous generation products because SOA is an approach, not a technology. (As Steve V. has pointed out at the end of this post it actually would make more sense to call it Service Oriented Approach so it isn’t confused with more precise software architectures such as hub and spoke, three tier, or REST.)
The comparison of the “traditional” vs the “incremental” approaches is difficult since everyone’s SOA requirements are different, and everyone’s budgets are very tight. It just seems more reasonable to offer a smaller priced offering for companies getting started with SOA, especially those sensitive to demonstrating ROI.
The point is that to do SOA you may not need a lot of software in the first place. You probably don’t need another application server, an EAI broker, or a message queuing system. The point of an SOA is to join these and other types of existing software systems together using a standard interface and some (possibly combination of) data formats and communications protocols.
So the incremental approach makes more sense for SOA infrastructure than it would for application servers, database management systems, message queuing software, packaged applications etc. since by definition what SOA is trying to do is put these kinds of things together, not replace them.
I am very pleased that the “Web of Services” workshop has been announced.
This workshop is open to W3C members and non-members. This workshop represents the next step in a somewhat lengthy discussion about the importance of Web services to enterprise software.
I am really looking forward to pinning down what, if anything, is wrong with Web services, and what, if anything, the W3C should do about it. Especially in the context of using Web services for SOA implementation.
Following the original submission of SOAP the W3C was the place for Web services specifications. This was the case until WS-Security was submitted to OASIS in September 2002. Since then Web services specifications work has been split among W3C, OASIS, WS-I, and the Microsoft workshop process.
One thing the industry lacks is a single point of leadership. Perhaps W3C should try to take that on? Or maybe OASIS would be a better place?
Another thing we tried for a couple of years was to create a Web services architecture. But we didn’t ever fully reach our goal. Don’t get me wrong, there’s a lot of good information in that document (and I know that the OASIS SOA RM used it), but it falls a little short of defining exactly what a Web services architecture is and how to tell whether or not a specification fits within it.
Other people often say that the REST approach is better than WS-* (usually described as “SOAP vs REST”) and this has pretty much been an endless argument in the blogosphere since it started (and was an endless argument at the WS-Arch WG as well). Maybe the workshop results can help resolve this, although that may be like suggesting the merger of two large established religions Anyway…
Does the Semantic Web have a role to play in the enterprise? Many would say “yes,” but this is another somewhat lengthy debate without much resolution. Maybe through the workshop we can identify enterprise software requirements and use cases that will settle this one, too…
The most important thing is to hear loudly and clearly from the users about their requirements for enterprise software standardization, especially for SOA based applications, and what needs to be done to truly achieve it.
I used to be a bit sceptical about Wikipedia degenerating into a forum for endless discussion, but now I’m a Believer and I use it all the time.
Not to mention the great entry on Robert Wyatt, whose version I really like (readers of this blog will not be surprised at that and Wyatting, of course… (can you believe that?)
ps I vote for keeping Wyatting as a separate entry
You know how we are always saying “the best technology doesn’t always win”?
And how someone always cites Beta vs V.H.S., and how Betamax was superior technology?
According to James Surowiecki in this week’s New Yorker that isn’t really the case.
Actually V.H.S. was the better product since it was a better match for what its users wanted to do:
Betamax fans still extoll its superior picture quality, but for most consumers V.H.S. was the better product; Betamax tapes could fit only an hour’s recording time, while V.H.S. could record an entire movie.
The article is actually about the current standards battle between Blu-ray vs HD-DVD.
But it also brings up the eternal question about how technology standards are agreed:
Yet standards themselves are determined in unstandardized ways.
And what can happen to resolve a war. In some cases, of course, multiple standards can successfully co-exist, such as Coke and Pepsi (he says) but in other cases the victory results in a “winner take all” situation such as Microsoft achieved with Word.
(By the way Surowiecki doesn’t predict a winner in the current war, saying both sides are doing things right – instead he implies negotiation will be required such as what happened with the DVD.)
The most encouraging thing about the article, however, it its perspective that even though consumer confidence is potentially the biggest factor (i.e. everyone wants to know they are investing in the likely winner), which results in a lot of vendor posturing, in the end the best solution usually wins.
Betamax was really not better than V.H.S., after all…
Not long after I posted about reuse I had the good fortune to visit a company embarking on an ambitious reuse program.
For obvious reasons of confidentiality I can’t mention the company name or give any significant details.
However what I can say is that it is a fairly large company with a large IT budget, and they have done a very careful study of reuse, including organizational issues, skill set issues, budgetary issues, and so on. They have executive buy-in. They even have a prioritized list of candidate services.
They realize it will be a challenging project but they are confident that it will improve their overall IT effeciency and reduce the overall IT spend.
So a little evidence on the other side of the question…
My daughter called last night and I was telling her that started this entry 4 days ago and still hadn’t managed to finish. She said, it doesn’t have to be perfect! Just go with what you’ve got… Finish it later. Just throw in a bunch of links and a short description
And so here it is…
Through Ivan Casanova I got invited to the Revision3 launch party last week. I could not believe I did not think to bring my camera, since I usually do. But I did find a detailed description of the party that includes a link to a Flikr photo page…
Before the party about all I knew what that getting your blog listed on Digg could really increase traffic. Ivan told me that Revision3 basically got started with Diggnation, which is sort of like a Wayne’s World based on the top items on Digg.com.
So it makes a lot of sense for Kevin Rose to get involved with Revision3 – a kind of natural link had been established between the two.
It was a lot of fun going, and I have downloaded a couple of the shows. They are definitely different and energetic – kind of video blogs or video podcasts. I am not sure that Internet TV is going to put the commercial networks or cable TV out of business any time soon, but it is great to have something new and different to check out.