it's keep going

More Share - Day 2
Spent the afternoon and evening getting lots of good info and talking to the folks at SHARE...
I found out a lot about the IBM stack, and how these guys think the pieces go together.
I got to talk to a lot of the Hursley types, and this has really helped me to understand how they see the pieces of WebSphere fit together. I have to say that all of the IBM guys seem sharp, approachable, and pretty honest about the ups and downs of IBM software.
Been Waitin for the ESBus
I think the whole idea of an Enterprise Service Bus is a bit of a hangup. To IBM, it is a glue between business components and back-end systems expressed as a set of design patterns. IBM's conception of the ESB seems to lie heavily on synchronous MQ-based messaging. On top of this, they speak of 'micro-flows', which are message processing and data transformation activities through their MQ Integration Broker software stack. Here is a level of data integration.
Layered on top of the ESB, IBM sees process flows, or 'macro-flows'. These seem to be based on traditional workflow ideas, and more recently, on BPEL4WS-type SOA coordination activities. Like most people, there is still this schizophrenia about web services. All the marketing hype from all of the vendors and talking heads is all about web services, but if you press the engineers, they will allow that they maintain an escape clause. Sometimes, web services just won't cut it. At any rate, your business processes, and your SOA orchestration activities, are run at this macro level.
I feel like I cleared up some misconceptions I had about their ESB view. I think terminology is a problem, as ESB and JBI move up the hype curve. Even the IBM guys admit that there is a lot of overlap in how these products are packaged and presented. It's not getting any easier.
Back on the Mainframe
Allright, the whole SHARE deal is a bunch of Z/OS guys, it's a mainframe/CICS conference at heart. You have to take everything with a grain of salt...same with NFJS, same with JavaOne...everyone has an axe to grind.
That said, I found a lot of customers (developers and sysadmins) who went with WebSphere on Z/OS, or who left Z/OS for distributed deployments of J2EE, and were coming back. I also popped into talk to some of the Z/OS JVM engineers, and got some more skinny on Java performance on Z/OS, and using the zAAP Java processor. I talked to folks who had done some pretty rigorous analysis for their shops. It does appear to be working for some.
CICS and the CICS TG
The CICS Transaction Gateway v6, and CICS TS 3.1 combo was a big item. After talking with folks, I really came away with the impression that J2EE -> JCA -> CICS TG is really the way to go to integrate CICS transactions into distributed applications.
CICS 3.1 was portrayed as a bit of a retrenchment, in a sense going back to traditional strengths. It sounded like JCICS and EJB within CICS has not taken off like originally conceived by IBM. On the other hand, there were customers that were beginning to use some of this, and were pretty happy with the results. It just seems like the emphasis has shifted more towards WebSphere and integration, with CICS being seen more as a tier removed.
It's not just us
I've done lots of Oracle 10G App Server lately, and coming back to WebSphere in catch-up mode. Like anything, it's a love/hate proposition. I was amused to hear some presentations from end-users that got into the difficulties of running app servers. It sounds like the WebSphere types wrestle with the same questions:
Problems with a lack of fine-grained access control to allow non-sysadmins to manage parts of the stack.
Problems with 'who does the deployments, and who fixes it when they go wrong'?
Problems with developers downloading code from the internet, and having a crisis when that code isn't compatable with production.
Problems with issue resolution...who does problem triage on distributed applications?
Problems with debugging.
Problems with constant maintenance patching, and the need to test and retest as new patches go up.
Really, it was nice to see we were not alone in having to sort all of this out, these issues were across the board with all of the folks I talked to.
A long day indeed...