240 likes | 246 Views
Introducing Model Driven System Engineering for Thales systems A Thales Unit Deployment Story. Agenda. What is MDSysE A Thales Unit Deployment Story Lessons learned Today’s Change Management Organization Future Change Management Organization. Agenda. What is MDSysE
E N D
Introducing Model Driven System Engineering for Thales systemsA Thales Unit Deployment Story Research & Technology
Agenda • What is MDSysE • A Thales Unit Deployment Story • Lessons learned • Today’s Change Management Organization • Future Change Management Organization Research & Technology
Agenda • What is MDSysE • A Thales Unit Deployment Story • Lessons learned • Today’s Change Management Organization • Future Change Management Organization Research & Technology
<<sys.Context>> Context <<sys.Logical>> Logical architecture <<sys.Physical>> <<sys.EPBS>> Physical architecture EPBS What is MDSysE? MDSysE: Tooling environment providing support for Model Driven System Engineering • MDSysE closely aligns with • SYS-EM Thales Methodology for system engineering • SysML concepts Standard for modeling systems. Formalization of system requirements Validation Product Verification Allocation of requirements Product Integration Formalization of component requirements Research & Technology
MDSysE’s IDE Research & Technology
Agenda • What is MDSysE • A Thales Unit Deployment Story • Lessons learned • Today’s Change Management Organization • Future Change Management Organization Research & Technology
A Thales Unit Deployment Story • First evaluation of MDSysE February 2005 • Assumption: process is self explained through the tool functions. • Results • MDSysE is a generic tool that needs to be specialized • MDSysE implements a process framework • Many process structure decisions need to be taken • E.g. What to refine and how refining from level of abstraction to level of abstraction is not clear. • There is a need to refine the method for the Thales unit. • Some defects needs to be fixed. Research & Technology
A Thales Unit Deployment Story • Second evaluation of MDSysE mid may 2005 • Delegate full time a TRT consultant to help the Thales Unit to elaborate its methodology. • 3 Thales Unit Process & tool experts involved in this task. • Objective: Build the method and qualify the tool for an effective deployment early 2006 • First step: mid may 2005: post mortem engineering of an existing system. • Try to emulate the mental process of system engineers and describe how the tool can support it. • Objective: Capture the decisions in slides that will serve for the creation of a “tool and process” training. • We do not target to create a model of the system. • Second step: October 2005: parallel engineering of a new system. • Do not impact the existing process. No synchronization constraints. • Capture with the new process what the engineering team has build. • Objective: Obtain the model of the system. • Complete and validate the process. Complete the training Slides. • January 2006: Engineer a new project with MDSysE. Research & Technology
Current Results • First Step has been covered • 230 Slides Produced to capture the considered Thales Unit process. • All the system levels of abstraction handled by MDSysE clarified for the Thales Unit • We need more generic behavior in MDSysE to build a process that can support more system engineering concerns. • Second Step Covered • Third Step started Research & Technology
Agenda • What is MDSysE • A Thales Unit Deployment Story • Lessons learned • Lack of definitions • Lack of best practices • Technology instability • Multiplicity of processes • Collaterals effects • External impacts • Cultural gap • Today’s Change Management Organization • Future Change Management Organization Research & Technology
Lessons learned (1) • Lack of guidance • What are system engineering flavors? • Software predominant systems. • Hardware predominant systems. • Object orientated systems. • We need to define guidance to select the process type at first or build the process before executing it. • What is system engineering? • Each Thales Unit has a different perception of what it « should » be. • Thales builds a wide wealth of systems. « System » means different things for different units. • Each Thales unit (project?) has a different perception on how a system should be elaborated. • Target objective not assigned to system engineering activities. • E.g. What doe the system engineer targets to check or define with the physical architecture description? Throughput? Distribution? Other? • No Thales Unit has time or is willing to harmonize its views with others. • Is this necessary? • What is system engineering versus software engineering? • When saying “this is no more system engineering but software engineering?” • To which level middleware has to be abstracted in the system architecture? • … Research & Technology
Lessons learned (2) • Lack of best practices • Example: What is the optimum level of granularity for a system capability? • How to measure cohesion, isolation, completeness etc. • What rules could help us to assess our work? • E.g. Use the 3rd normal form to assess the robustness of our data exchange model. • We need to capture and disseminate the system engineering best practices among Thales Units Research & Technology
Lessons Learned (3) • Technology instability • PIM / PSM does not mean anything • System externally visible behavior can be platform specific • There is still a lot to discover • How expressing the binding of an architecture on an execution platform? • How to guide architects in choosing and expressing their decisions? • Typically when binding an architecture on an execution platform • What tooling support can be provided to support product line engineering? Research & Technology
Distribution Internet Hard Real Time Time Distributed Soft Real Time No constraint MDSysE Embedded Culture Functional Decomposition Object Orientated Component based Lessons Learned (4) • Multiplicity of processes • The definition of an MDE/MDA process depends on multidimensional parameters. • E.g. culture • Functional decomposition does exists • Object orientated • Component based architecture Research & Technology
Lessons Learned (5) • Collateral effects • Deploying an MDA/MDE approach pulls other disciplines in the picture. In Thales case: • Revisiting the requirements taxonomy is a must. • Concept of “Capability” is too wide. It does not align with the semantic of a MDSysE Capability. • Integration and Test • Formal system modeling helps at mastering the integration process. • Test • Not yet addressed but System Test Cases link to system architecture. How? • Configuration Management: • System modeling provides the mean to build the configuration management strategy. • Revisiting the configuration management elements is a must • CSCI must be able to be broken down in smaller configuration items. This is not yet the case. • Software Engineering • Integration of system engineering modeling platform and Middleware design studio: • Technical integration • Semantic alignment MDE / MDA requires that there is some effort put in updating the corporate artifacts templates (SSS, SSDD, SRS etc.) Research & Technology
T0 SSS SSDD Time T0 SSS SSDD Time Lessons Learned (6) • Customer Relationship Short term Impact. • Deliverable milestones. • The set of documents to be delivered to the customer at well defined milestones may slip. • The system organization was not captured in UML models • The system organization is now captured in UML Model during the engineering process, but later. • The deliverables are now partly made of UML Model views. • Less ambiguity • More formality, consistency, completeness. • Does the customer understand this formalism? Does he agree the process? Standard scenario MDE / MDA scenario Research & Technology
Time Lessons Learned (7) • Cultural Gap needs to be managed • There is a lot to learn by the process end users • Need to build a phased coaching strategy to manage the cultural gap • Therefore for a Thales business line the deployment process depends on • The unit maturity • The technology progress • The level of dissemination. Maturity Dissemination Time Technology Time Research & Technology
Consequences • Create Specialized versions of MDSysE according to the needs of each unit. • Manage MDSysE as a product line. • Core set of standard modules • Domain specific modules • Need: Create a strong CCB process to help harmonizing / factorizing views. • Views cannot be 100% unique. • Put as much as possible in MDSysE Core. • Need: Create a strong coaching infrastructure acting as Thales units proxies. • MDSysE evolutions derive from: • Thales units needs. • Technology progress • The coaching infrastructure • Provides expertise in formalizing Thales know how in an MDE/MDA process. • Increase the shared part between Thales Units. • Capture and disseminate System Engineering best practices. • Are not tool experts. • Are not a specific business domain experts. Research & Technology
Agenda • What is MDSysE • A Thales Unit Deployment Story • Lessons learned • Today’s Change Management Organization • Future Change Management Organization Research & Technology
MDSysE4Unit B MDSysE4Unit A Today’s Management Organization Thales Unit MDSysE Software Development Team Coach CCB Deployment support • TRT Software Development team is in charge of developing the core of MDSysE. This excludes process specific behavior. • TRT Parameterization team is in charge of parameterizing MDSysE according to Thales Units process specific needs. • TRT Deployment support is in charge of educating each Thales Unit and guarantees the homogeneity and optimum reuse accross the Thales Units processes. Thales Unit Coach Parameterization Team Research & Technology
Agenda • What is MDSysE • A Thales Unit Deployment Story • Lessons learned • Today’s Change Management Organization • Future Change Management Organization Research & Technology
Need for an evolving organization • This organization will evolve • TRT cannot have the responsibility to maintain all the MDSysE flavors. • TRT needs to concentrate on common needs: • Need for provision of a pattern engine. • Need for provision of product line approach. • Need to provide a support for trade-off analysis. • … • TRT needs to provide the support required to easily create specialized versions of MDSysE. • MDSysE as a software factory. • Thales units must identify in their organization people in charge of maintaining their own instance of MDSysE. • Next level of Thales unit maturity. • Coaching is still required to facilitate understanding and formalization of units needs. Research & Technology
MDSysE Core Specialization Engine MDSysE Software Development Team Coach CCB Deployment support Coach Thales Unit Thales Unit Parameterization Team Parameterization Team Medium term organization target • This organization is already proposed for some projects Research & Technology
Question? Research & Technology