In the afternoon session of the 5th MDA and Agile Modeling Forum I got the following messages:
Fabrizio Gheller (Altran)
Need and advntages of applying a stability index to business objects to obtain better overall performances of processes.
Pierfranco Ferronato (Soluta.net)
Wrong perspectives on agile development:
- Agile as an excuse for avoiding planning and documentation
- Programers are the only knowledgeable of what happens at the business level
- Resource lock-in: if that guy leaves, we are dead. He is the only one who knows!
Sometimes there’s the need of joining business and IT, sometimes of splitting them: IT people should not talk business, and business ones should not talk IT .
UML is not a DSL, it’s a NDL: No Domain Language.Spaghetti modelling is not much better than spaghetti programming, and a tool or standard method per se will not solve your problems.
The problem should be split in:
- configure the business [IT independent]: parameters and success metrics
- model the business [IT independent]: steps and activities
- support the business [IT dependent]: new services and semantics
The developer will not write specific code any more, he will write templates.
Roberto Pozzi (IBM Rational)
The usual risk is separation between models and implementations. Diversified tools and languages are needed for allowing communication between business and IT and between planning and solution delivery, down to alignment of the infrastructures that support business services.
Rational System Architect is a toolsuite that provides different modeling views to the enterprise architecture design.
Objective criteria must be defined to keep activities aligned with the objectives. Risk analysis and metrics should be applied to model driven design too, to allow governance of the portfolio of projects.
Eventually, a unified and unique repository should be there for allowing search and reuse of all the assets. Metamodels should guide their classification (as from the “Rational Common Modeling Platform” vision).
IBM is not yet at modeling execution, but for sure model simulation is fully achieved.
Paolo Miola (Tibco)
He discussed the lessons learned from applying Tibco and bp reengineering to a large-scale bpm and bam scenario for Crédit Lyonnais (LCL) part of the Crédit Agricole Group (6,000 backoffice users and 60,000 documents managed daily).
- First phase: application of BPM and BAM to the scenario
- Second phase: applying rule-based dynamic job assignment based on workload, context and expertise of the various executors (geographically distributed)
The objectives are:
- process visibility: understanding and coding the processes
- process governance and unified management
- resource optimization (in the example scenario: 75% of human resources were spared and reallocated to more core activities)
The next wave for Tibco is predictive behaviour of systems: the two-second advantage: “a little bit of information beforehand is more valuable than all of the information after the event”.
Ernesto Damiani (UniMi)
Interestingly enough, on a set of 200 companies in Lombardy, a lot claim the have or want to have BPM solutions and BPM capabilities, but then 60% of them declare a time-to-implementation of more than 6 months for any BP change (= no flexibility) and 80% declare that no measure of productivity improvement after bpm automation is monitored.
Furthermore, typically, IT division and CIOs have no role in BPM (not only at the modeling level, but also at the procurement and technical level for enactment of models). Effect is that IT performs only support activity to modeling, and costs are allocated to the IT division, while decision is driven by high management.