Copyright 2001 Software Productivity Consortium NFP Inc

9ae8d2386710e7115d1bbe296585e9e4.ppt

Количество слайдов: 26

Copyright © 2001, Software Productivity Consortium NFP, Inc. SOFTWAR E PRODUCTIVITY CONSORTIUM Integrating Information Models: the essential element of successful measurement programs COCOMO Forum October 2004 Chris Miller Software Productivity Consortium

Copyright © 2004, Software Productivity Consortium NFP, Inc. 2 Integrating Information Models Motivation • Popular emphasis on goal-based measurement, but — Anyone can have goals — Goals often are not consistent — Organizational goals may be independent of projectspecific issues • Measurement initiatives often implemented as disjoint collections of data • Measurement systems across the enterprise must be aligned to provide a true enterprise perspective • Quantification of casual relationships provide the basis for predictive analysis and forecasting performance

Copyright © 2004, Software Productivity Consortium NFP, Inc. 3 Integrating Information Models Information Needs Customer Satisfaction Enterprise TIME-TOMARKET Process Project QUALITY Market INNOVATION Product Sales PRICE

Copyright © 2004, Software Productivity Consortium NFP, Inc. 4 Integrating Information Models Types of Information Needs Enterprise Measurement should support business goals and objectives Information needs flow downward Process Business performance Efficiency and effectiveness of production Project Accomplishment of project budget and schedule Product Satisfaction of customer requirements Data and information are passed upward

Copyright © 2004, Software Productivity Consortium NFP, Inc. 5 Integrating Information Models Enterprise View • Need to measure business performance — Focusing on strategic goals — Reporting profitability (program, business unit) — Providing information for marketing • Enterprise may ask questions such as: — Should we invest in a new product line? • When present, an enterprise policy for managing and measuring drive measurement behavior and actions

Copyright © 2004, Software Productivity Consortium NFP, Inc. 6 Integrating Information Models Balanced Scorecard *Source: Robert S. Kaplan, and David P. Norton. “The Balanced Scorecard – Measures that Drive Performance, ” Harvard Business Review, January 1992.

Copyright © 2004, Software Productivity Consortium NFP, Inc. 7 Integrating Information Models Process View • Concurrent processes exist in a typical enterprise (systems, acquisition, development, operations, logistics, marketing, …) • Focus on the process capability — Definition of processes – Including measurement points and vehicles — Adherence to defined processes – Self-audits — Improvement of processes – Effectiveness – Efficiency

Copyright © 2004, Software Productivity Consortium NFP, Inc. 8 Integrating Information Models Product View • The system itself is the primary product — Component quality characteristics — Need to assess impact of changes across the system components — Customer satisfaction with product • Products and deliverables — System architecture, specifications, software, interface control documents, test cases & procedures, users’ guides, …

Copyright © 2004, Software Productivity Consortium NFP, Inc. 9 Integrating Information Models …Applied Across the Life Cycle Critical process performance measured in process, during process execution Maintenance Phases Defects Development Phases Product quality characteristics measured and addressed at product transition points throughout the life cycle

Copyright © 2004, Software Productivity Consortium NFP, Inc. 10 Integrating Information Models Project Views • Focus on the activities and tasks as a project — Meeting product delivery dates — Planning effort — Monitoring and controlling effort – Adherence to management plan — Customer satisfaction with performance

Copyright © 2004, Software Productivity Consortium NFP, Inc. 12 Integrating Information Models Integration of Views Common Process Tailored Support Senior Mgmt Commitment Level-Specific Information Process Info. Models Project Info. Needs Systems Software Product Info. Needs Project Info. Models Product Info. Models Automate Evaluate Process Info. Needs Analyze Systems Eng. Software Eng. Enterprise Info. Models Plan Enterprise Info. Needs

Copyright © 2004, Software Productivity Consortium NFP, Inc. Integrating Information Models 13 Integration Activities 1. Each view maintains it’s own set information models for monitoring their own information needs 2. Senior management commitment — Buy-in is needed at the most senior level to facilitate and drive an integrated measurement program (after all they will benefit the most) 3. Use common base measures — When planning to do measurement use a standard — Standard consists of common ways to analyze and evaluate the data and measurement process 4. Build automated tool support around the standard base measures and analysis techniques — Requires tailoring for each view

Copyright © 2004, Software Productivity Consortium NFP, Inc. 14 Integrating Information Models Step 1: Information Models by View Measurement Information Model • Defines measurement product • Links measurable entities to information needs • Provides rigorous definitions of basic concepts • Measurement construct consists of : — Base Measures — Derived Measures — Indicators Information Needs Information Product Measurable Concept Measurement Construct Entities Attribute Information Model • Explains prevalence of three-level measurement models (e. g. , Goal-Question-Metric [GQM])

Copyright © 2004, Software Productivity Consortium NFP, Inc. 15 Integrating Information Models Measurement Process Model Technical & Management Processes INFORMATION NEEDS USER FEEDBACK ANALYSIS RESULTS Core Measurement Process Database Establish Commitment Plan MEAS UREMENT PLAN Perform Evaluate ANALYSIS RESULTS AND PERFORMANCE MEASURES Experience Base IMPROVEMENT ACTIONS Adapted from ISO/IEC CD 15939, Information Technology – Software Measurement Process

Copyright © 2004, Software Productivity Consortium NFP, Inc. 16 Integrating Information Models Step 2: Buy-in with Business Case Executive Management Project Management Business Goals Engineering Management SPI Activities Image Retention Referrals Award fees Bonuses Market Share Profit Customer Satisfaction Fewer Failures Shorter schedules Lower Price Higher Margin Low-cost provider Time-tomarket Lower support costs Cost Reduced Effort Quality Reduced Rework Cycle Time Increased Capacity Productivity

Copyright © 2004, Software Productivity Consortium NFP, Inc. 17 Integrating Information Models Map Goals/Measures/Activities (Cycle Time Example) Business Goals, Objectives Engineering Analysis, Response Better Estimates Quality Reduced Labor (Cost) Early to Market Shorter Schedules (Cust. Sat. ) Reduced Rework Cycle Time Increased Capacity Productivity Predictable Schedules Estimating Model Requirements Management Change Control Stable Requirements Efficient Process Historical Data Fewer Steps, Delays Process Mapping Process Engineering Desired Results of Activities Defined Process Measurement Program Defined Process Initiatives

Copyright © 2004, Software Productivity Consortium NFP, Inc. 18 Integrating Information Models Step 2 Exit Criteria • Business case needs to: — Obtain senior management commitment — Show information flow down — Highlight key activities • Provide business-oriented information for each view — Potentially new information needs — Highlight linkage to other views

Copyright © 2004, Software Productivity Consortium NFP, Inc. Integrating Information Models 19 Step 3: Unify Definitions • Base measures need to remain consistent across all measurement activities: — Uniform base measures — Data formats — Identify common entities and attributes • Analysis techniques are tailored to generate unique indicator base on individual information needs: — Define derived measures — Consistent analysis techniques — Identify common reporting standards and graphic for related indicators

Copyright © 2004, Software Productivity Consortium NFP, Inc. 20 Integrating Information Models Uniform Base Measures • Consistent use of — Data labels (e. g. , SLOC) — Data formats • Base measures are the lowest common denominator — Validate across all information models • Avoid redundant data collection methods • Maintain integrity of scales and units Base Measure Definition/Template Base Measure Name Release 01. 03 Lines of code Attribute C++ language statements Measurement Method Count semicolons in all program units Type of Measurement Method Objective Scale (including precision) Whole numbers from 0 to infinity Type of Scale Ratio Unit of Measure Line of code

Copyright © 2004, Software Productivity Consortium NFP, Inc. 21 Integrating Information Models Uniform Analysis • Mitigate redundant naming conventions between indicators — (e. g. , Productivity) • Set standard analysis approach, techniques, and methodologies • Ensure decision criteria in consistent across information models (and views) • Use consistent graphic representations for similar data Indicator Definition/Template ID Cost Performance Index for Project XX Information Need Enter ID of information need Measures Used Derived Measure(s) Cost Performance Index (CPI) Base Measures 1. Actual Cost of Work Performed 2. Budgeted Cost of Work Performed Function(s) Calculate CPI: For any given project, perform the following calculation: Actual Cost of Work Performed (ACWP) CPI = Budgeted Cost of Work Performed (BCWP) Analysis Model Simple CPI calculation. Optimal performance is to keep actuals close to or at the center line value of 1. 0. Decision Criteria Expected value is ‘ 1. 0’. Decision needed if actual value is +/– 10%. Graphical Representation. Standard CPI line graph with value of 1 represented as a horizontal line, positive values above and negative values below.

Copyright © 2004, Software Productivity Consortium NFP, Inc. 22 Integrating Information Models Step 4: Tool Support • Automated tools support is required to handle the mass of data collection and analysis associated with an integrated measurement program across many views • Base measure definitions and analysis techniques provide a basis for selection of tool suite • Understanding the information flow in the business case helps during trade-off decisions among different tools

Copyright © 2004, Software Productivity Consortium NFP, Inc. 23 Integrating Information Models Summary • An effectively designed measurement program must consider the interrelationships of the views • An integrated measurement program will need to establish: — Uniform base measures — Common tools and databases • Each view uses the base measures to: — Define indicators • Allows for the flexibility for each view to address their information needs • Integration is an impossible task without senior management support

Copyright © 2004, Software Productivity Consortium NFP, Inc. 25 Integrating Information Models Measurement Construct Information Product Information Needs Indicator Estimate or evaluation that provides a basis for decision making Model Interpretation Algorithm combining measures and decision criteria Derived Measure Value resulting from applying the algorithm to two or more measures Algorithm combining two or more base measures Function Base Measure Value resulting from applying the method to one attribute Method Operations mapping an attribute to a scale Attribute Property relevant to information needs Source: Adapted from ISO/IEC 15939, Software Measurement Process Framework Entities

Copyright © 2004, Software Productivity Consortium NFP, Inc. 26 Integrating Information Models Measurement Construct Example Systems Engineering performance – documentation quality? Information Needs Indicator summarized for all SE documents in monthly status Information Product Compare defect density Interpretation to organizational average, perform causal analysis if > 1. 0 defects per page 0. 5 > Organization average of 0. 3 defects per page Indicator Model Compare defect density to SE organizational data Derived Measure Function Derived Measure Divide (defects per page) 50 defects found Base Measure 100 pages reviewed Count only bullets in minutes SE Entities Document Peer Review Base Measure Method Count only technical pages Defects per peer review Attribute Pages reviewed Source: Adapted from ISO/IEC 15939, Software Measurement Process Framework