Agile adoption in a waterfall environment
5 likes1,070 views
The document discusses the transition from a traditional waterfall model to an agile approach in a large call center application development project led by Valtech. Key improvements included shortened development cycles, increased collaboration, and a focus on reducing defects through practices like automated testing and daily stand-up meetings, resulting in significant enhancements in delivery quality and team communication. The agile evolution was achieved with management support and a focus on adapting processes without imposing drastic changes.
Agile adoption in a waterfall environment
- 1. AGILE ADOPTION IN A WATERFALL ENVIRONMENT BY KEVIN MURRAY PRINCIPAL CONSULTANT // KEVIN.MURRAY@VALTECH.CO.UK // TWITTER: @KEV_C_MURRAY
- 2. BACKGROUND TO PROJECT VERY LARGE CALL CENTRE CMS FOR JOBCENTRE PLUS SEEN AS ONE-STOP-SHOP FOR CLAIMING BENEFITS 30,000 USERS, UP TO 5,000 CONCURRENT USERS HIGHLY COMPLEX APPLICATION MULTIPLE SYSTEMS INTEGRATORS (VERY POLITICAL LANDSCAPE) COMPLEX INTEROPERABILITY WATERFALL SOFTWARE DELIVERY LIFECYCLE UN-DYNAMIC CUSTOMER
- 3. VERY TRADITIONAL WATERFALL 12 WEEKS REQUIREMENTS 12 WEEKS DEVELOPMENT 12 WEEKS SYSTEM TEST 12 WEEKS UAT 48 WEEKS TO THE END OF UAT
- 4. BACKGROUND TO PROJECT VALTECH BECAME DEVELOPMENT LEAD IN 2005 INTRODUCING RUP AND MAKING IMPROVEMENTS STRAIGHT AWAY INTRODUCED ITERATIVE DEVELOPMENT INTRODUCED PRIORITISATION BASED ON TECHNICAL COMPLEXITY STARTED TO INTRODUCE AUTOMATED UNIT TESTING AND CONTINUOUS INTEGRATION
- 5. ITERATIVE DEVELOPMENT 4 x 4 WEEKS WEEKS REQUIREMENTS 4 x 4 WEEKS DEVELOPMENT 4 x 4 WEEKS SYSTEM TEST 4 x 4 WEEKS WEEKSUAT 28 weeks to the end of UAT (plus 25% more dev)
- 6. KEY PROJECT ISSUES LAST MINUTE RUSH AT THE END OF EVERY ITERATION SYSTEM TEST WERE A SEPARATE TOWER, LOOKING FOR EXCUSES TO SLIP HIGH NUMBER OF DEFECTS HIGH DEFECT RATES IN SYSTEM TEST AND UAT QUALITY WAS STILL BETTER THAN PREVIOUS DEVELOPMENT LEAD MOST DEFECTS NOT SEEN AS A DEVELOPMENT TEAM PROBLEM! COSMETIC DEFECTS, EXTERNAL DESIGNS CREATED IN VISIO AMBIGUOUS REQUIREMENTS CURRENT LIVE PROBLEMS BLAME WAS FAILING BETWEEN SILOS
- 7. COST OF DELIVERY PAIN DEFECT TOTALS INTO 3 FIGURES FOR EACH RELEASE AFTER FIRST ITERATION, 25% OF DEV EFFORT WAS SET ASIDE TO FIX DEFECTS TEAM OF 20, EQUATED TO APPROXIMATELY 300 MAN DAYS JUST FIXING DEFECTS DURING THE DEVELOPMENT CYCLE (OUT OF 1600 DEVELOPMENT DAYS PER RELEASE)
- 8. WHY WE WENT AGILE AWARENESS OF DELIVERY PAIN ALWAYS LOOKING TO IMPROVE WE WERE BECOMING VERY AGILE AWARE WE HAD A CATALYST, AN AGILE COACH HAD JUST JOINED THE TEAM
- 9. MANAGEMENT AND CUSTOMER BUY IN VALTECH LED TEAM HAD MADE SIGNIFICANT IMPROVEMENTS PROGRAMME AND PROJECT MANAGERS HAD COMPLETE TRUST AND ALLOWED US TO TAKE THIS FORWARD END CUSTOMER WAS AWARE BUT WASN’T TO BOTHERED AS THERE WASN’T A DIRECT IMPACT ON THEM
- 10. AGILE EVOLUTION
- 11. AGILE EVOLUTION – ADDRESS LAST MINUTE RUSH DAILY STAND UP MEETINGS CONSCIOUS DECISION NOT TO CALL IT A SCRUM USED ROTATING SCRUM MASTER, LARGE TEAM, EVERYONE GOT INVOLVED KEPT TO 15 – 20 MINUTES PER DAY DAILY STAND UP MEETING ALWAYS HAPPENED AT THE SAME TIME PROVIDED ACCURATE STATUS UPDATES, DAILY MOTIVATED THE WHOLE TEAM ADDRESS ISSUES EARLIER
- 12. AGILE EVOLUTION – ADDRESS LAST MINUTE RUSH BURNDOWN CHARTS EARLY VISIBILITY PRINTED OUT AFTER STAND UP & STUCK ON THE WALL BY THE OFFICE DOOR TRUST AND TRANSPARENCY - PROJECT AND PROGRAMME MANAGERS SAW DAILY PROGRESS
- 13. AGILE EVOLUTION – OUR BURNDOWN
- 14. AGILE EVOLUTION – OUR BURNDOWN
- 15. AGILE EVOLUTION – ADDRESS HIGH DEFECT RATES RELEASE AND ITERATION PLANNING ALL DEVELOPERS GOT THE BIG PICTURE, NOT JUST TEAM LEADERS VERY INCLUSIVE, WE ENSURED BA’S, TESTERS AND EVEN THE DEPLOYMENT TEAM ATTENDED DEVELOPER AND BA COLLABORATION REAL SCREEN SHOTS’ IN EXTERNAL DESIGN DOCUMENTS
- 16. AGILE EVOLUTION – ADDRESS HIGH DEFECT RATES FORMAL EMPHASIS ON AUTOMATED UNIT TESTING (DEFINITION OF DONE) DEMONSTRATIONS RETROSPECTIVES OVERHEAD OF GETTING EVERYONE TOGETHER AND RELUCTANCE FROM THE LARGE TEAM TO CONTRIBUTE SATISFACTORY FEEDBACK ACQUIRED FROM DAILY STAND UPS
- 17. AGILE EVOLUTION – WHAT WE DID NOT DO. ESTIMATING ALREADY HAD ACCURATE MAN DAY ESTIMATING PROCESS REDUCE SPRINT SIZE SOME TASKS WERE TOO LARGE SPRINT PLANNING OVERHEAD SYSTEM TEST WOULD NOT BE ABLE TO COPE BUILD AND DEPLOYMENT OVERHEAD DUE TO TECHNOLOGY STACK QUESTION OR SACRIFICE EXISTING GOVERNANCE AGILE ACTUALLY AIDED THE WATERFALL GOVERNANCE NOT WHOLESALE CHANGES AND WE DIDN’T RISK DELIVERY
- 18. BENEFITS NO END OF SPRINT PANICS USING MS PROJECT YOU DON’T ALWAYS GET THE FULL PICTURE COMPARED TO STAND UPS. IF SOMETHING WASN’T GOING TO BE DELIVERED WE GOT KUDOS TELLING EARLY, RATHER THAN ON DELIVERY DAY BETTER QUALITY REDUCED DEFECTS TO SINGLE FIGURES TECHNICAL DEBT PROPERLY ADDRESSED CODE QUALITY TEST COVERAGE AGILE WAS SPREADING SILOS WERE COMING DOWN
- 19. BENEFITS MORE COLLABORATIVE RELATIONSHIP WITH BAS WE IDENTIFIED BUSINESS BENEFITS WHERE IT WAS SUPPOSED TO BE AN ISSUE WE IDENTIFIED AND ADDRESSED A BA BOTTLENECK DEVELOPERS COULD INTERCHANGE AS BAS WE STARTED TO FOCUS ON BUSINESS CRITICALITY PROJECT CULTURE CHANGE NEW INTEGRATION PROJECTS HIT THE GROUND RUNNING PEOPLE, PROCESSES & INFRASTRUCTURE ALL AGILE
- 20. PROJECT OUTCOME NOT WHOLESALE CHANGE ADAPT AND EVOLVE, DON’T JUST ADOPT MANAGEMENT SPONSORSHIP AND TRUST UNDERSTAND YOUR BOUNDARIES CHALLENGE WHERE YOU CAN DOCUMENTATION GOVERNANCE VISIBILITY, SO AGILE SPREADS ORGANICALLY LEARN FROM MISTAKES CAPTURE KEY INFORMATION
- 21. NO NEGATIVE NEWS HEADLINES // SYSTEM COPED DURING THE RECESSION LEGACY