Distributed Scrum Fast, Lean, and Scalable Agile Project Management with Outsourced Development Teams Jeff Sutherland, Ph. D. Co-Creator of the Scrum Development Process http: //jeffsutherland. com/scrum worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Jeff Sutherland jeffsutherland. com/scrum n Agile Systems Architect since 1986 – CTO/VP Engineering for 9 software companies – Prototyped Scrum in 4 companies – Conceived and executed first Scrum at Easel Corp. in 1993. Rolled out Scrum in next 5 companies – Scrum consultant to leading companies in Europe, North and South America, and Russia. – Senior Advisor, Open. View Investments, LLC n Signatory of Agile Manifesto and founder of Agile Alliance n Scrum Certification World Tour Feb 1 -2 CSM Philips Feb 7 -8 CSM Aarhus Feb 22 -23 CSM Ft Myers Mar 1 -2 CSM St. Petersburg March 12 -13 CSM QCon London March 22 -23 CSM Boston Apr 3 -4 CSM Amsterdam Apr 11 -12 Scrum Tuning Boulder Apr 18 -19 CSM Boston Apr 25 -26 CSM Aarhus Apr 28 -29 Deep Agile Boston with Ron Jeffries May 7 -11 Scrum Gathering Portland May 17 -18 CSM Boston May 31 – Jun 1 CSM Stockholm Jun 6 -7 CSM Aarhus Jun 11 -12 CSM Copenhagen Jun 13 Oresund Agile Copenhagen Jun 14 -15 Saxo Bank, Copenhagen Jun 18 -19 CSM Oslo Jun 20 -21 CSM Stockholm Jun 23 -24 St. Petersburg, Russia Jun 26 -27 CSM Amsterdam July 1 -2 CSM Rome worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Scrum World Tour Produced by Trifork – www. trifork. com – contact Anne Sofie Bille asb@trifork. com worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Are you doing Scrum? The Nokia Test by Bas Vodde n First, you must be doing iterative development – Iterations must be timeboxed to less than six weeks – Software must be tested and working at the end of an iteration – Iteration must start before specification is complete n Then you must meet the Nokia Scrum test n 1969 - Earliest published reference to Iterative Incremental development – Robert Glass. Elementary Level Discussion of Compiler/Interpreter Writing. ACM Computing Surveys, Mar 1969 – See Larman, Craig and Basili, Vic. Iterative and Incremental Development: A Brief History. IEEE Computer, June 2003 (Vol. 36, No. 6) pp. 47 -56 worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
For those of you doing Scrum n You know who the product owner is n There is a product backlog prioritized by business value n The product backlog is has estimates created by the team n The team generates burndown charts and knows their velocity n There are no project managers (or anyone else) disrupting the work of the team Kniberg, Henrik. Scrum and XP from the Trenches: How We Do Scrum. Version 2. 1, Crisp, 5 Apr 2007. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Experiences with people doing Scrum Hyperproductive Scrum Teams n It is easy to double productivity with Scrum. Just implement the basics. Focus on implementing only high business value features. n To quadruple productivity requires surfacing impediments and removing them (inspecting and adapting). This can produce Toyota level productivity and quality. n Scrum was designed for more, i. e. 5 -10 times productivity improvement. This has been experienced in three types of teams: – The first Scrum team and similar colocated teams. – The first distributed Scrum team. – The Sirsi. Dynix distributed/outsourced teams. n Understanding hyperproductive Scrum teams can help leverage your distributed software development with Scrum. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Putnam Process Productivity Index Industry Average n N = 1. 27*(N-1) Systematic IDX Primavera n 33 is 2098 times 1 Easel product Sirsi. Dynix Easel development Mike Cohn Scrum Borland Quattro “There are software development organizations that are more productive than others. ‘Luck’ is not the whole story; you need to look for ways to improve. ” Myers 1998 Industrial Strength Software: Effective Management Using Measurement by Lawrence H. Putnam and Ware Myers (1997) worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Hyperproductivity in Distributed Scrum n IDX Systems – 567 developers, dozens of locations n IDX Web Team 1996 -2000 – Burlington, VT – Boston – Seattle n Factors accelerating the IDX hyperproductive Distributed Scrum – Scrum organizational pattern – Engineering practices – Daily meeting of distributed team – Tools (direct connection to Microsoft development) worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Distributed Scrum Styles Isolated Scrums Distributed Scrum of Scrums Totally Integrated Scrums worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Outsourcing n Outsource $2 M development n Outsourcing costs - $1. 6 M – Industry data show 20% cost savings on average n Introduce Scrum locally – 240% improvement at IDX, for example n Local Scrum costs – $0. 83 M n Sirsi. Dynix radically reduced outsource costs making oursourcing reasonable for: – Gaining expertise that is unavailable locally – Expanding and contracting development staff without layoffs worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Jack Blount CTO Sirsi. Dynix n 30 years in the computer software industry n Began as a software engineer in Raleigh, NC for IBM n SVP Engineering at Novell in the 80 s n CTO at Tele. Computing, Mobile. Ware, JD Edwards n COO at Borland, Raindance n CEO at USDA, Picus, Mobile. Ware, Dynix, Alpha. Bay Nick Puntikov VP Exigen Services (formerly CEO, Star. Soft) n USSR Academy of Sciences n XP Advocate worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Overview of 4 th Generation ILS – 8. 0 n ILS is like a ERP system for a vertical market n 3 rd Generation ILS was a client/server app running at more than 12, 000 locations in 42 countries n Supports access by more than 120 million people a year n 4 th Generation is a completely new development effort – 100% Java development – More than 1 million lines of source code – More than 3 thousand features – Kerberos security – Record ownership – Consortium support worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Decision Criteria to Outsource n Fixed R&D budget based on declining sales of old product n Limited Java expertise available locally n Critical time to market to achieve $30 million growth in sales based on stalled market and sales pipeline for new product n Dramatically improved performance, security and scalability required by market worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Outsourcing Selection Process n Researched outsourcing firms in India, China and Russia – Internet search – Personal network n Why these countries – India well recognized as the outsourcing leader with largest number of companies and engineers – China attempting to position itself as the next India for outsourcing, plus we have an office and large customers in China – Russia while a small outsourcing industry compared to India, Russia has long led the world in mathematics worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Outsourcing Selection Process n Contacted multiple firms for phone interviews – 18 in India – 7 in China – 9 in Russia n Scheduled on-site interviews – 6 firms in India – 4 firms in China – 5 firms in Russia n Personally flew to all three countries and held the interviews from both a technical and business perspective worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Outsourcing Selection Process n Key criteria for decision – Reputation – References – Java experience and skills – System software experience and skills – Development processes in place – Development tools used – Ability to staff project – Business process for contract negotiation worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Risk Associated with Outsourcing n Unable to staff project on time n Unable to keep staff on the project once it starts – India and China have about 35% annual staff rotation – This is the generally viewed as the largest exposure and always results in delayed projects n Poor communications – Remote development always has communications challenges – Different time zones of US and outsource countries – Lack of formal processes – Language barriers worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Project Plan For Success n Detailed selection process led to the RIGHT outsourcer n Tight implementation of SCRUM development process n Two week Sprints n Daily code check in by all developers at all locations n Used shared source code repository with US developers n Integrated outsource team into core team n On site upfront training in Russia n Daily scheduled SCRUM calls – 15 minutes n Bi-weekly software demonstrations as combined team worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Sirsi. Dynix Distributed Scrum n Over a million lines of Java code worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Sirsi. Dynix Distributed Scrum n 56 developers distributed across sites PO PO PO Sirsi. Dynix Provo, Utah SM Dev Dev T Ld Dev Dev Catalogue Exigen Services St. Petersburg, Russia Serials Circulation Search worldwide scrum consulting practice Reporting © Jeff Sutherland 1993 -2006
Sirsi. Dynix Distributed Scrum n Scrum daily meetings St. Petersburg, Russia 17: 45 pm Local Team Meeting 7: 45 am Provo, Utah Scrum Team Meeting worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Sirsi. Dynix Distributed Scrum n Common tools worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Sirsi. Dynix Distributed Scrum n Uncommon performance Colocated Scrum* Waterfall* Sirsi. Dynix Distributed Scrum** Person Months 54 540 827 Lines of Java 51, 000 58000 671, 688 Function Points 959 900 12673 FP per dev/month 17. 8 2. 0 15. 3 *M. Cohn, User Stories Applied for Agile Development. Addison-Wesley, 2004 **J. Sutherland, A. Viktorov, J. Blount, and N. Puntikov, "Distributed Scrum: Agile Project Management with Outsourced Development Teams, " in HICSS'40, Hawaii International Conference on Software Systems, Big Island, Hawaii, 2007. . worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Conclusion – Jack Blount n Project was on schedule every step of the way n One year project resulted in a 50 person engineering team creating more than 1 million lines of Java code – 24 engineers in Russia – 26 engineers in US n Project has been recognized as the largest most successful outsourced SCRUM Java project in the world n Product is in final Beta and customers are very excited by new features, new look and feel, new performance and new security worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Experience Summary – Jack Blount n Previous outsourcing projects in India resulted in very limited success – Primary problems were constantly re-staffing project – Engineering talent and processes in India more tuned toward maintenance and recoding legacy products rather than design and implementation of new solutions n Star Software is an outstanding Agile development partner – – Excellent communications Excellent processes Excellent talent Russian engineers on project rated as top performers in overall annual appraisal process worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Linear Scalability of Scrum Projects Scrum Teams Velocity Waterfall Project Size • J. Sutherland, A. Viktorov, J. Blount, and N. Puntikov, "Distributed Scrum: Agile Project Management with Outsourced Development Teams, " in HICSS'40, Hawaii International Conference on Software Systems, Big Island, Hawaii, 2007. • J. Sutherland, C. Jacobson, and K. Johnson, "Scrum and CMMI Level 5: A Magic Potion for Code Warriors!, " in Agile 2007, Washington, D. C. , 2007. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Poppendieck Lean Thinking Tools P 1 Eliminate waste P 2 Amplify Learning P 3 Responsible decisions P 4 Fast Delivery P 5 Empower team P 6 Buil integrity in P 7 See the Whole Tool 1: Eliminate Waste Tool 3: Feedback Tool 7: Options Thinking Tool 10: Pull Tool 13: Selfdeterminatoion Tool 17: Perceived integritet Tool 21: Measures Tool 2: Value Stream Mappig Tool 4: Iterations Tool 8: Latest Responsible Moment Tool 11: Queue Theory Tool 14: Motivation Tool 18: Conceptual Integritet Tool 22: Contracts Tool 5: Synchronization Tool 9: Decision Making Tool 12: Cost of Dealy Tool 15: Leadership Tool 19: Refactoring Tool 16: Expertise Tool 20: Test Tool 6: Setbased development worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Systematic’s new model for Lean SW development Value P 6 Build Integrity in Engineering T 19 Refactoring T 20 Test Flow Pull Perfection P 2 Amplify learning P 2 Amplify Learning P 6 Build Integrity In T 5 Synchronization T 4 Iterations T 3 Feedback T 6 Setbased development P 1 Eliminate Waste T 2 Valuestreams T 11 Queuing Theory T 22 Contracts T 12 Cost of Delay T 21 Measures T 10 Pull P 5 Empower team T 16 Expertise Management P 4 Fast Delivery P 5 Empowerteam Tools can be divided in three dimensions P 7 See the whole T 14 Motivation T 15 Leadership T 18 Conceptual integrity T 17 Perceived integrity P 3 Decide in latest Responsible moment T 7 Options thinking T 8 Latest responsible Moment T 9 Beslutningstagning P 5 Empower team People T 13 Self-determination These are thinking tools – Projects and employees knows best how to transform them worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Systematic Pilot Projects n Selected projects were asked if they would like to pilot improved processes related to test and reduced cycle time respectively. n Project staff were trained in the Lean mindset and asked to suggest how to adopt Lean into their processes. n The result was a selection of Scrum and early testing based on story-based development. The pilots were planned and completed. n The result of the pilots were two-fold: it confirmed the general idea of using Lean mindset as source for identification of new improvements, and secondly it provided two specific successful improvements showing how agile methods can be adopted while maintaining CMMI compliance. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Systematic Pilot – Small Project n First pilot was initiated on a request for proposal – Systematic inspired by Lean principles suggested a delivery plan with bi-weekly deliveries – Stated explicit expectations to customer involvement and feedback. – The project had a teamsize of 4 and concerned software for a customer in Danish Government. n Key reasons for Systematic award: – commitment to deliver working code bi-weekly – providing a very transparent process to the customer. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Small Project Success Factors n Delivery plan and customer involvement resulted in early detection of technology issues. – Had a traditional approach been used these issues would have been identified much later with negative impacts on cost and schedule performance. n Productivity of small project was at the expected level compared to the productivity performance baseline for small projects. n Another small project with a team size of 5 working for a Defense customer using Scrum shows a similar productivity and the same indicators of high quality and customer satisfaction. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Pilot of Larger Project n Team of 10 worked on a military messaging system. – This project was inspired from the Lean thinking tool “Build Integrity In” to investigate how to do early test, and as a result they invented a story based approach to early testing in software development. – The name “Story based” development was inspired from XP, but the approach included new aspects like: short incremental contributions, inspections and was feature driven. n The idea of story-based development was to subdivide features of work, typically estimated to hundreds of hours of work into smaller stories of 20 -40 hours of work. – The implementation of a story followed a new procedure, where the first activity would be to decide how the story could be tested before any code was written. – This test could then be used as the exit criteria for implementation of the story. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
New Approach to Testing Reduced Defects by 38% n Many benefits from story-based development were immediately apparent. The combination of a good definition of when a story was complete, and early incremental testing of the features, provided a very precise overview of status and progress for both team and other stakeholders. n Developing a series of small stories rather than parts of a big feature is more satisfactory, and creates a better focus on completing a feature until it fulfills all the criteria for being “done”. n This project finished early, and reduced the number of coding defects in final test by 38% compared to previous processes. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Another Larger Project n Team of 19 working on a module to a electronic patient record system, also worked with early testing. n They ensured that test activities were integrated into development, with a strong focus on “seeing the whole” and understanding how the solution fit into the customers domain. n For each week the project defined a goal to be achieved. The project ensured that test and domain specialists were co-located with the developers. This caused discussion and reflection between testers, developers, user experience engineers and software architects, before or very early in the development of new functionality. n As a consequence the amount of remaining coding defects in final test were reduced by 42% compared to previous processes. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Conclusions from Larger Projects n Test activities should be an integrated activity through out the projects lifetime, and Scrum inherently supports this, through cross-functional teams and frequent deliveries to the customer. n Story-based software development method should be the default recommended method for software development in projects. n Test activities should be an integrated activity through out the projects lifetime worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Systematic adoption of Scrum in at CMMI Level 5 n Process Action Teams (PATs) were formed to integrate the experience and knowledge gained from the pilots, into the processes shared by all projects in the organization. n The largest change to project planning is that features and work are planned in sufficient detail as opposed to a complete initial detailed analysis. – Result is a Scrum Product Backlog with a complete prioritized list of features/work for the project. – All features have a qualified estimate, established with a documented process and through the use of historical data, but the granularity of the features increase as the priority falls. – The uncertainty that remains is handled through risk management activities. n The primary change to project execution processes, is to integrate Scrum as method for completing small iterations (Sprints), on a selected subset of the work with highest priority. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Published experiences with ”rework” Part of development time 50% ~50% 45% 40% 35% 30% 25% ~25% 20% 15% ~15% 10% ~10% 5% CMMI 1 CMMI 2 CMMI 3 Source: Krasner & Houston, Cross. Talk, Nov 1998 Diaz & King, Cross. Talk, Mar 2002 worldwide scrum consulting practice CMMI 4 ~7% CMMI 5 © Jeff Sutherland 1993 -2006
SCRUM and PDP-Common worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Control chart of build process Build errors Unit test errors FXCop failed worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Rework at Systematic 12% 10% 9, 8% 8% 8, 3% 6, 9% 6% 6, 4% 6, 0% 7, 6% 4% 6, 8% 4, 7% 2% Q 2 2005 Q 3 2005 Q 4 2005 Q 1 2006 Q 2 2006 Q 3 2006 worldwide scrum consulting practice Q 4 2006 Q 1 2007 © Jeff Sutherland 1993 -2006
Scrum applied to CMMI Level 5 company – 6 month results Project effort 100% Rework 100 % Work 90% 80% 50 % 70% Process focus CMMI 69 % 9% 60% SCRUM a 50% 35 % 4% 40% 30% 20% 50 % 25 % 10 % CMMI 1 6% CMMI 5 SCRUM worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Systematic CMMI 5 Analysis First six months of Scrum n 80% reduction in planning and documentation costs (still under discussion) n 40% reduction in defects n 50% reduction in rework n 100% increase in overall productivity n Systematic decided to change CMMI Level 5 process to make Scrum the default mode of project management n When waterfall project management is required, they are now need to be contracted for twice the price of Scrum projects – – – Required by some defense and healthcare agencies Results are lower business value Lower customer satisfaction Lower quality Twice the cost Sutherland, J. , C. Jacobson, et al. (2007). Scrum and CMMI Level 5: A Magic Potion for Code Warriors! Agile 2007, Washington, D. C. , IEEE. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Sutherland, J. , A. Viktorov, and J. Blount. Adaptive Engineering of Large Software Projects with Distributed/Outsourced Teams. In Proceedings of the International Conference on Complex Systems. 2006. Boston, MA, USA. worldwide scrum consulting practice © Jeff Sutherland 1993 -2006
Скачать презентацию Distributed Scrum Fast Lean and Scalable Agile Project
0a05e7d094b734e4a469ff587d2f71b5.ppt