Agenda THR%%A%P COM 1?4w 3 About Xerox 3 Lifecycle Anxiety Waves at Xerox 3 Spiral Case Study 3 Life After Spiral... 3 Anchored Software Process 3 Anchored Lifecycle Macro Process 3 Technology Readiness 3 Lifecycle Selection Caveats 3 Market-driven Product Development Slide 2
- Copier Company? Think Again... TII~~M-I' COMPOY,IFF DIR 4 r=@ 3 Lifecycle Anxiety Waves at Xerox 'r~ ~%r?l&fih (.XMllANI Pre- 1987: No lifecycle awareness Waterfall IS the lifecycle model. ~1987-92: "What should we do with the waterfall?' Standard, documented processes are still waterfall, in some cases mimicking existing hardware processes. Improvised solutions to mitigate waterfall shortcomings. -1993-95: "Let's try the spiral!" Managers of a major program plan and operate around the spiral. Software Risk Evaluation in '95 shows discouraging results. ~1996-97: "What should we do? Fallback to waterfall improvisations. Random exposure to Objectory, Rational processes and tools. = 1997-99: Adapting the Anchor Point concept Push for incrementaviterative processes and platform development for variant products. Y2000 challenge: Product Line Processes Slide 4
I IIKWM~I. CO~IIWY Case Study Program Characteristicis Major, strategic program Open, scaleable, high performance core for a family of printers (50-600 pagelmin, including color) Large number of PDLs Enterprise-wide connectivity Open, industry standard platforms (SPARC first) Distributed 0-0 Design and Development Spiral is the pivotal development model Slide 5 Program Plan Characteristics?'~~%~fih'r COM I?~XY - -- -- -- -- - Slide 6 Planned 6 turns to generate the first major release Planned 6 Architecture Definition Iterations Out of the 21 major line-items of the plan: (4) Software related 4 review milestones are planned for software (2) Hardware related Drawing, and Builds (2 Engineering and 1 Manufacturing). Hardware design overlaps feature specifications and the first iteration of the software architecture definition (1) System Integration and Test 6 integration periods are planned ( 14) Non-engineering aspects Marketing, Customer engagement, pricing, sales training, etc.
Spiral Model* m?i%y&%.r COM PAW Objectives, Alternatives A /- Evaluate Alternatives RA: Risk Assessment Plan next phases Slide 7 * Based on Boehm, Barry W: "A spiral model of software development and enhancement," IEEE Computer, 1988 Issues identified '~'HKWW@%~ COM1'4N' Slide 8 During the phases not only the prototypes became "throwaways", but the architecture versions as well Risk analysis was superficial, and also inefficient As a result, architecture never stabilized Overly aggressive plan created an overload of new technologies Technology experimentation obfuscated architecture development Resolution of technology risks was overwhelming, further preventing the stabilization of the architecture
Slide 9 Life after Spiral... *M~;~W~KM.I. COMI~ANY "Spiral", as a powerful metaphor and graphic representation of returning and repetitive activities is here to stay The key, new concept is anchoring* the process We need to address the issue of synchronizing many spirals for complex systems We also need to address the synchronization of software spirals with hardware waterfalls** * Boehm, Barry W: "Anchoring the Soflware Process," 1EEE Software, Volume 13, No 4, July 1996, pp 73-82 ** Rechtin, Eberhardt and Maier, Mark W., "The Art of Systems Architecting,", CRC Press, 1997, pp 56-60 Anchored "Lifecycle Macro Process" This is the recommended solution to ease the synchronization of many spirals, and the synchronization of spiral software development with the hardware waterfall. "Technology Readiness" Rigorous technology acquisition, delivery and demonstration process Slide 10
Anchored Software Process 1-1 I R%S{%%.~. COM ~%YY XCROX LC0 LCA IOC (SDR) (EOM) 9 9 Anchor Point Definitions per Boehm: LC0 Life Cycle Objectives LCA Life Cycle Architecture IOC Initial Operational Capability New, Xerox-specific Anchor Points: SDR Software Delivery Readiness EOM End Of Maintenance Slide 11 I Generic rrproduct" (ResearchlQQ Hardware
Slide 15 Lifecycle Selection Research Highly creative, inventive process. Most of the time runs on open-ended schedule. Lack of result is acceptable. Researchers most of the time are not very receptive to any procesdlifecycle ideas, but spiral model is appropriate. Technology Definition (or Acquisition) The risk still exists that the application of research results or the use of available off-the-shelf technology is not feasible or viable. The spiral model is appropriate, since the considered risks are primarily technical. Simply put, many times the risk assessment will simply conclude that "... it will not work under the circumstances." Technology Readiness It needs to be demonstrated that the hardware will be manufacturable, and neither the hardware nor the software will need extraordinary, open-ended efforts during the development and manufacturing process phases. 1 TH Lifecycle Architecting Caveats ~FYfitKfi51' COM PAW It is 0.k. to combine Research and Technology It is not 0.k. to combine Research or Technology Development with Product Development I Note that non-engineering risks, such as business risks are not comprehended in this model. Slide 16
Business DomainIMarket Risks I Phase-gate reviews A A A A A ~nchor Point reviews O a HWISW Engineering Domain Risks Business and engineering domains represent different kind of risks, consequently different risk mitigation approaches needed. Engineering life-cycle models must have strong architecture focus. I Slide 17 Any Questions? THE 'r COMIWY Slide 18 Your head is spinning?