The Rise & Fall(?) of Modelling

Transcription

1 The Rise & Fall(?) of Modelling MARK THOMAS UK LEAD SW ARCHITECT, THALES UK Ver

2 Contents The need for models The Hype Curve The Rise - Thales experience The Fall - The Challenges New Opportunities Q&A 2

3 The Need for models Simplify Complexity Improve Understanding Explore Architecture Alternatives / Trade-offs Define / Quantify subject matter Visualise perspectives and change Share/Capture Knowledge Communicate with Stakeholders Store Corporate Knowledge Create Documentation Enable Reuse Faster, Cost Effective, High Quality, Product Line Engineering Realise Automation Validation Simulation / Demonstration Generation 3

4 The Hype Curve What is it? A representation of the maturity, adoption and social application of specific technologies. Modelling - Where are we today? Let s try something new - UML It works great oh? Aha Ooh, something new! - SysML Do we ever get to the Plateau? e? The Hype Curve (Gartner) - 4

5 Some of Thales Peak Experiences (UK) Generic Vehicle Architecture (GVA) Model Based Software Deployment on the Foxhound vehicle Evolutions towards wider LOSA Submarine Sonar Model Based Software Development on UK Sonar systems Code Generation and Deployment of all Signal Processing code Infra Red Threat Warner (IRTWS) Model Based Infrastructure for Automatic Coding and Deployment of Matlab Algorithms on target 5 Others

6 The Disillusionment Why isn t everyone modelling? For every success there have been various failed attempts! Excuses - Our needs are too different! - It is too expensive! Lack of realised Benefit - Projects focus on the adoption issues rather than the long term benefits! Reluctance to Change - If it worked last time, can we just do the same again? 6

7 The Challenges to achieving the Plateau Complexity, What Complexity? Systems often begin simple and well-understood and then through natural evolution become more and more complex. Developers fail to perceive a need until much too late. Open Source software SW modelling is no longer the focus for developers Engineers comfortable exploiting reusable code-based assets Agile adoption Implies short code development/delivery cycles Developers are comfortable sticking to their integrated development environment (IDE) and debugger No time to Model!, Just work more hours! 7 Culture Traditional programmers are very proficient at conventional coding techniques. Modelling requires additional training and tools, a corresponding investment in time, money and effort is needed! Traditional developers believe modelling slows them down.

8 Some evidence! Recruitment agencies no longer focusing on UML (it s all about Agile and languages) MDA re-titled by the OMG to mean Model-Driven Advantage Engineers are not demanding access to the latest modelling tools! CodeGen Conference is taking a fallow year in 2015 and the focus in the last few years has moved to DSLs/DSMLs, certainly not UML! Open source UML tooling (under eclipse) still immature in terms of supporting the standard (and that s after many years of development effort) Open source executable, translatable UML tooling not capturing the imagination of developers since its release by Mentor Undergrads not using models (other than as informal graphics) hence have little interest when seeking jobs in industry 8 Large software developers not looking for modelling expertise (e.g. CGI, Google)

9 But we are not looking at just one curve? Complex systems require a wide variety of Modelling technologies Formal v Informal Models Physical v Abstract Models - Descriptive Models - Analytical Models - Hybrid Models Domain Models Simulation Models - Executable Models (over Time) and others We need them all to work together! 9

10 There are triggers (new opportunities) happening all the time! Feature Modelling and Variability 10 A key enabler for Product Lines DSLs/DSMLs Formal Methods To solve the really hard problems! xuml and Domain Modelling Tools Capella is now available Open Source We need to fit them all together Increased Interoperability Open Services for Lifecycle Collaboration

11 Physics Maths Networks CAD And we are not looking at just one problem! As Software fights its way out of disillusionment Systems is discovering modelling. Enterprise Level Architecting (The AF s ) System Level Architecting (SysML, ) - A curve already well travelled by SW! Enterprise System Software 11 We need them all to work together!

12 Back to the Curve We are continuously progressing along the curve, and triggering new opportunities! Modelling approaches are simultaneously rising and falling across the Business! Work still needs to take place on The pieces (FM, DM, xuml, ) The Interoperability of the pieces The modelling story continues! 12

13 Thank You 13

14 The Rise and Fall(?) Of Modelling Abstract Thales considers model-based engineering a good practice, and continues to invest in the development of modelling methods & tools to support the engineering function. However, the complexion of modelling use across our business has changed. Whilst earlier focus lay in the traditional areas of software and algorithm development, our attention has turned to supporting co-engineering - creating continuity between modelled systems and software solutions. Likewise, as our business moves in the direction of Product Line Engineering, we see a rise in the production of Feature Models to support Variability Management and Enterprise Models to help understand Need. By contrast, our software community is today thriving on the availability of large OSS code bases, requiring little or no reliance on modelling expertise. The ebb and flow of modelling seems to be the norm today, with its form and purpose adapting to meet our business challenges. This presentation highlights some of the variety of key modelling trends today in Thales, and illustrates how and why modelling practices are variously in the ascendency and in decline. 14