Modeling Complex Systems at IBM Research

1. Modeling Complex Systems at IBM Research August 12 2008, Dagstuhl MECS Workshop IBM Research, Alan Hartman, Sugata Ghosal (IRL) Ali Arsanjani, and others (GBS, WRL) Hiro Akatsu, Hiroshi Ishikawa (TRL) Julia Rubin, Andrei Kirshin, Dolev Dotan, Dany Moshkovich (HRL)

2. Four Technologies from IBM Research for dealing with Complex Systems • Model Driven Systems Engineering • System Grokking • Service Oriented Modeling and Architecture – Modeling Environment • Product Line Support IBM CONFIDENTIAL

3. Requirementdocument analysis Requirement modeling Construction Test Analysis and design Modeling Support Infrastructurefacilitates systems engineering from the view point of method and process, modeling environment, and team development Modeling Support Infrastructure Rational Method Composer Method and process • Method and process • Work guidance based on development method • Process definition and control with existing project management system Guidance personalization and navigation Team development Modeling environment Rational Modeling Platform • Modeling environment • MDSE method based systems modeling with modeling tools. Process Management System Model Repository Model-based product information traceability • Team development • Model artifact management in team development • Mapping and traceability among models and other artifacts Tabular expression of models for analysis

4. Validating System Models • Continuous behavior is modeled in Simulink • Discrete Behavior modeled in UML • SysML profile is used to model the System • MARTE profile is used to define timing constraints • UML model references Simulink modules • Simultaneous integrated simulation of UML and Simulink Simulink UML Execution Engine Integrated Simulation SysML MARTE Discrete behavior Continuous behavior

5. R R Model analysis and verificationverifies and analyzes combination of discrete and continuous models for product specifications with time and parametric constraints • Architecture – Components for Rational Modeling Platform Co-simulation control for Simulink TPTP-based test driver Time-aware UML execution engine Execution Trace Analyzer Timed test cases Timed discrete behavior Continuous behavior Virtual time scheduler Test result with Performance metrics Time constraints (OMGTM MARTE profile) Parametric constraints (OMGTM SysML) • Test results: • Expected parameter values • Metrics: • Throughput • Response time • Resource utilization Parametric constraint evaluator Time constraint evaluator * Simulink is a registered trademark of The MathWorks, Inc. * OMGTM is a trade mark of the Object Management Group

6. Four Technologies from IBM Research for dealing with Complex Systems • Model Driven Systems Engineering • System Grokking • Service Oriented Modeling and Architecture – Modeling Environment • Product Line Support

7. C++ C Java High level RE Models Introduction PROBLEM Low level RE Models Legacy code • Systems are large, complex and semantically opaque • Existing knowledge described in form of low level elements: Existing RE tools don’t raise the level of abstraction Packages Interfaces Classes Structures Procedures Statements Expressions … System Grokker SOLUTION • Raise the level of abstraction by discovering: Layers Components Services Features Capsules High level behaviors Patterns … • Add semantic description

8. Technology • System Grokker - tool that supports incremental and iterative user driven system understanding, validation and evolution • Provide framework for manipulation and visualization of various groups of domain specific software elements and relationships between them. It also supports: • Addition of semantic attributes to any element in the system • Marking elements with user defined tags • Union, intersection and substitution operations on groups • Supports semi-automatic discovery of domain specific high-level structural and behavioral abstractions, and other useful for modernization process groups of elements and relationships • Represent groups of results created by Rational Software Analyzer • Represent groups of domain elements matching search/query criteria • Provide additional analyzes on top of Rational Software Analyzer framework e.g. componentization, layering • Can be extended/customized for different domains

9. depends High Level Architecture Domain Domain Extensions Development Environment Java Domain to Semantic Model Transformations Semantic Model provides a common representation of information required by analyses on top of different domains and is used for visualization and persistency of analyses results C/C++ Cobol Analyses framework Domain specific analyses UML General analyses … Other technologies (RSAR, SUPA, ADERET, OCL, …)

10. Development environment Semantic Model Explorer Semantic Diagram Semantic Model Semantic Group Semantic Diagram Semantic Palette Semantic Group Semantic Adapter Semantic Adapter Semantic Attribute Visual Cue Semantic Relation Semantic Relation Semantic Properties

11. Examples • Pattern detection and validation example • Technology based system understanding • Package based layering example • Group by Feature example • Component identification example • Simulation of architectural changes example • Dependency reduction example

12. Pattern detection and validation example 1. The pattern consist of factories, interfaces, implementations, and the rest of the system elements. 2. Group elements that play the same role in different instances of the pattern. 3. Analysis of dependencies between main elements in the patterns reveals problematic relationships (in red). 4. Choose problematic relationships on the diagram (System  Implementation) and expose its sources for examination.

13. Component identification example • System Grokker can be used to identify a group of architecturally and semantically coupled elements that are recommended to be placed in the same component • For Object Oriented systems like Java and C++ the tool can be used to group tightly related Classes • For procedural oriented systems like C the tool can be used to group tightly related procedures, global variables and even attributes • User can customize and apply various decomposition heuristics and compare provided alternatives • User may modify provided solutions and repeat the decomposition process Variant A Variant B

14. Four Technologies from IBM Research for dealing with Complex Systems • Model Driven Systems Engineering • System Grokking • Service Oriented Modeling and Architecture – Modeling Environment • Product Line Support

15. SOMA Overview • Service-Oriented Modeling and Architecture (SOMA) is a modeling and design method aimed at enabling target business processes through the identification, specification, and realization of business-aligned services that form the Service-Oriented Architecture (SOA) foundation. • The SOMA method provides in-depth guidance on how to move from the business models created through the IBM Component Business Modeling (CBM), Websphere Business Modeler (WBM) or similar business analysis techniques and tools, to the IT models required by an SOA. • SOMA leverages the IBM Global Services method (The Method) execution models, work product descriptions (WPDs) and techniques.Other methods like RUP or client methods can also be leveraged.

16. Component Business Modeling (CBM) • CBM is a business consultancy method used by IBM Global Business Services • CBM is a technique to decompose an Enterprise into its constituent building blocks (business components) • A component is a logical grouping of people, technology, and resources that delivers specific business value, and can operate independently • The output of a CBM engagement (heat map, process models) can be the input to an SOA engagement • CBM tool is now part of WBM

17. Marketing & Customer Mgt. Products Rentals Management Rental Fleet Logistics Business Administration Direct Customer Segmentation Fleet Strategy Corporate / LOB Strategy Rental Product Strategy Location & Channel Strategy Product Development / Design Fleet Planning Customer Relationship Strategy Financial Management & Planning Location Design & Layout OEM Relationship Planning Channel Design & Layout Real Estate Planning Marketing Strategy & Planning Promotions Management Channel & Location Profitability OEM Performance Management Alliance Management Customer Behavior Modeling Pricing Management In-bound Logistics Control Business Performance Reporting Location Operations Management Market & Competitor Research Reservations Management Legal & Regulatory Compliance Segmentation Management Call Center Workforce Management Real Estate & Construction Management Campaign Management Risk Management Stock Ledger HR Management (Career Dev., Training, Recruiting) Execute Rentals & Reservations Location Operations Purchasing / Sourcing Customer Service HR Administration / Payroll Fleet Servicing Demand Forecasting Time & Attendance Preferred Member Mgmt Corporate Audit Fleet Management Corporate Accounting (GL, AP, A/R, Treasury, etc.) Customer Communications Mass Marketing & Advertising Indirect Procurement Target Marketing PR & Investor Relations IT Systems & Operations CBM Heat Map of Rent-A-Car “Hot” Components E X A M P L E For illustration only

18. Rent-A-Car Process Process Model 0.RentVehicle 1.1 Reserve Vehicle 1.2 Check out Vehicle 1.3 Check in Vehicle E X A M P L E For illustration only 1.1.1 Check Rates 1.1.2 Make Reservation 1.2.1 Locate Reservation 1.2.2 Modify Reservation 1.2.3 Create Rental Agreement 1.2.4 Sign out Vehicle from Lot 1.3.1 Locate Rental Agreement 1.3.2 Process Return Information 1.3.3 Process Payment 1.3.4 Return Vehicle to Lot 1.1.1.1 Get Location (pick-up/drop-off) 1.1.1.2 Get Date / Time (pick-up/drop-off) 1.1.1.3 Choose Vehicle 1.1.1.4 Get Options Information 1.1.1.5 Check Vehicle Availability 1.1.1.6 Offer Rates for Selection 1.1.2.1 Confirm Rental Information 1.1.2.2 Get Customer Information 1.1.2.3 Get Payment Information 1.1.2.4 Confirm Reservation 1.1.2.5 Create Reservation Heat map and process models are used as inputs to SOMA

19. SOMA Modeling Environment - An Introduction • Adds SOMA support to Rational Software Architect and Rational Software Modeler. • Based on CBM SOMA v3.1 and SOA Solution Stack v1.27. • Extends UML meta-model for SOMA methodology - UML 2.0 Profiles for SOMA. • Provides Model Template suitable for SOMA phases and activities. • Automates repeatedly performed tasks with MDD enhancements for SOMA - transformations and patterns. • Ease modeling with User Interfaces to create SOA/SOMA model elements. • Model Validation for completeness and consistency. • Reduces effort and duration drastically with SOMA Work Product Generation capabilities.

20. SOMA-ME: Accelerating SOMA Adoption and SOA Development Labor Acceleration RSA/RSM 7 SOA Patterns & Transformations • Automatic Generation of Deliverables • Normative Guidance and automation for model creation • Model Validation • Definition and Generation of Downstream Work Packets SOMA Steps Enablers Work Product Generation Enhanced SLT Lifecycle Acceleration SOMA Process Guidance & Artifact Capture GSM WorkProducts SCA/WBSFCode Artifacts

21. SOMA-ME Roll Out and Field Interactions • Results in 2007 • Four client engagements have used the tool successfully • Other account teams are evaluating the tool and are waiting on licensing decision • GBSC has used SOMA-ME in 11 projects that span 6 industries • 430 practitioners have downloaded the tool • 182 practitioners have been trained in using the tool (hands-on) • 258 more have participated to training workshops (SOMA-ME overview/demo) • Ongoing Activities in 2008 • TLE 2008 SOMA-ME Labs (2) • IBM SOA Community of Practice (CoP) Announcement and Lecture Series on SOMA-ME 3 • F2F/Online Training Workshops • DeveloperWorks article series on SOMA-ME planned. Research and GBSC Taiwan preparing articles on CBS development using SOMA-ME • Field Engagements – GBSC, Client Engagements • GBSC Adoption Path – Roll out for 800 Practitioners

22. Four Technologies from IBM Research for dealing with Complex Systems • Model Driven Systems Engineering • System Grokking • Service Oriented Modeling and Architecture – Modeling Environment • Product Line Support

23. Case Study - Samsung’s Digital Printer Platform (Product Line) INKJET MONO LASER MFP INKJET COLOR LASER MFP MONO LASER MFP COLOR LASER 59 Printers 23

24. Design - Architecture of Printer Platform DPSVC Printer Services Service Manager PictBridge SmarThru JMS Smart Panel PrintSS ScanSS CopySS CSS Printer Middleware ImSS CommSS FaxSS OSAL HAL Toner Ink No-Nois ScanLens Memory HDD Power Paper VxWorks pSOS Embedded Linux x86 MIPS ARM Albatross ATI Etc Base Platform 24

25. Architecture of the Printer Platform in RSA

26. Architecture of the Main Printer Component in RSA Variability Support

27. From Platform Model to Products – Samsung Digital Printers Platform Model (Product Line Model) Components + Variability Update,Add New Features RunValidation Configuration ProductSCX-4321Model Product MJC-8700Model Any other printer… Transformation Transformation ProductSCX-4321Build Scripts ProductMJC-8700Build Scripts