How a large German Transport company uses Function Points for Estimating Dominique Lafourcade QuantiMetrics GmbH Gustav-Stresemann-Ring 1, 65189 Wiesbaden, Germany
Agenda Introduction The company & it s figures History & evolution of metrics, FP s and Estimating within the company FP Governance: how is the FP counting organized? Projects: How are FP s used in estimating Benchmarking: what role does it play 2
Introduction My story, our company, our customers (briefly) Why present a positive example Large companies, big problems, complex solutions 3
Three main activity areas: people transport goods transport The company & it s figures main branches: transport network maintenance & enhancement Some key figures: 2.7 Billion travelers per year 341 Million tons transported per year 63 914 km dedicated network to maintain 5 707 dedicated charge discharge terminals 2009 Revenue: 29 335 Million Euros 4
Over 500 systems supported The company & it s figures : the IT Department 240 000 man days effort expended in application development during 2007 71 000 Desktops 1,2 Petabyte Disk Storage / 3,2 Petabyte Backup capacity (1 PB is approx 1 M GB) 300 Million visitors on Internet Portal during 2008 Main systems supported: Passenger Reservations (quote some volumes ) POS ticket terminals (145 M tickets in 2008) Logistics Scheduling 5
IT Delivery is organized by Functional areas Functional Areas People Transport Goods Transport Logistics -Project -Functional -Business -Project -Functional -Business -Project -Functional -Business Process Management Guidelines, Repository, Benchmarking, Gate reviews Platform 1 Technical Technical Developer s Technical Resource pool Platform 2 Technical Technical Developer s Testing -Test 6
Key Success Factors in the Introduction of Functional Metrics Focus on the simple implementation Have one steering group for Functional Counting Gather count results into a centralized repository Comparison amongst groups should have a limited audience Counting and data gathering should take place in the functional team Effort expended in counts should be commensurate with the value of the results Reduce the counting effort by using average values ( Rapid ) Training and Communication Training at different levels and depths (management, analysts, technicians, counters) Correct common misconceptions through accurate and recurrent communications Functional metrics are used as one of parameters in estimating, however an estimating tool/method needs to be used in order to produce an estimate 7
Step by step introduction of Functional Measurements 2005 2006 2007 Metrics for Software Development launched Selection of Function Points as metric of choice Integration of FP in the development process Training of Function Point counters Set up of FP database FPA and Cocomo adopted as a standard Expanded FP training for, Project leaders and Business analysts Establishment of FPA by Business area 2008 Full integration in the existing processes and toolkit 2009 Introduction of the concept of FP delivery rather than Man Time Quotations delivered in Functionality rather than Effort 8
History & evolution of metrics, FP s and Estimating within the company People Transport -Project -Functional -Business Functional Areas Goods Transport -Project -Functional -Business Logistics -Project -Functional -Business Process Management Guidelines, Repository, Benchmarking, Gate reviews Function Point Counters Platform 1 -Technical -Technical -Impl. Technical Areas Platform 2 -Technical -Technical -Impl. Process Management Testing -Test Function Point Steering 9
Projects: How are FP s used in estimating The Cocomo Model calibration: how the estimate is broken down Cost: effort/fp Time : time/fp Quality : time, scope, budget, defects Define Project Functional Requirements Technical Design Implement Test Deploy Maintain As agreed 16% of total 13% of total 40% of total 15% of total 5% of total % of total Gate review Requires FP count and Cocomo Estimate FP Baseline count FP enhancement counts per release Estimates pre RFQ to support negotiations 10
Resources Probability Projects: How are FP s used in estimating Cocomo model: choices to be made 50% versus 90% Realistic Estimate 50% Secure Estimate 90% Secure planning might lead to blocked resources Secure Estimate Effort Realistic Estimate Time 11
History & evolution of metrics, FP s and Estimating within the company Frame contracts with IT suppliers now (since 2009) include Function point productivity requirements Estimating for projects is done using multiple techniques and merging them: Avoid common sense errors that may result from a model Use available experience & resources Provide more accuracy in the estimate 12
Benchmarking: what role does it play Productivity -Compare yourself to the market to set your targets -Track your performance year on year Quality Speed My Company Comparison group median Comparison group best 25 % 13
Questions? 14
Many thanks for your attention! 15