Cost/Performance Roadmap SRR Issue Description Cost/Performance Analysis Approach Baseline Cost Allocation (Preface to Tradeoffs) Cost Tradeoff Analysis ¥ Operations Staffing Options ¥ Evaluation of increased processing/storage requirements of Òtall-poleÓ products ¥ Processing vs Storage Cost/Performance Issues Heard at SRR SRR presented cost drivers with no relative cost sensitivity information Cost models too conservative about the likely advances in computing technology Costs seem too sensitive to floating-point operations required for product production Cost constraints are more likely in populations of users and the rates that they may be served Need to know relative cost of various functions to make priority tradeoffs Cost/Performance Analysis Cost/Performance Roadmap Cost Allocation Cost allocations based upon historical models, modified for evolutionary development, and intensive negotiations ¥ Evolutionary development - Prototyping effort - Out-year allocation to accommodate change Presented as aid in doing cost sensitivity analysis Cost Allocation of Negotiated Baseline (Total Program) Total Program = $766M Cost Allocation Backup Information (Total Program) Out-Year Allocation to accommodate change ¥ Effort allocated post 98 ¥ Includes Science Office, engineering, developmentÿeffort to evolve system based on user feedback, technology enhancements ¥ Includes amount for COTS evolution Engineering Studies ¥ Effort used to support special studies, analysis and development ¥ Data migration support ¥ Allocation on task direction from ESDIS Universityÿ Research ¥ Educational research for prototyping in areas of user interfaces, data access and management and other key technologies ¥ University research in area of software reuse processes Cost Allocation of Negotiated Baseline (Development Labor) Development Labor = 27 % of Total Program = 27% of $766M = $207M Cost Allocation Backup Information (Development Allocation) System Engineering ¥ Requirements / external interface analysis ¥ System architecture/design/operations concept ¥ System modelling ¥ Life cycle cost analysis ¥ DAAC engineering liaisonsÿ Segment Development (SDPS, CSMS, FOS) ¥ Segment engineering ¥ Design, code and checkout ¥ Segment integration & test Science Office ¥ Science user interface for requirements understanding ¥ Algorithm Integration & Test Support ¥ DAAC science liaisonsÿ Cost Allocation of Negotiated Baseline (M&O) M&O = 34% of Total Program = 34% of $766M = $260M Cost Allocation of Negotiated Baseline (DAAC Operations Support) DAAC Operations Support = 74% of M&O = 74% of $260M = $192M DAAC Staffing Examples Cost Allocation of Negotiated Baseline (COTS by Type) COTS = 14% of Total Program = 14% of $766M = $107M Cost Allocation of Negotiated Baseline (COTS by Component) Cost/Performance Roadmap Reduced DAAC Operations Options Baseline: ¥ Three 8 hr Shifts, 7 days/week (full functionality at all times) Alternatives: ¥ Two Shifts (prime time = 16 hours/day, 7 days/week) ¥ Extended Day Shift (prime time = 12 hours/day, 7 days/week) ¥ Day Shift (prime time = 8 hours/day, 5 days/week) Options during Ònon prime timeÓ: ¥ Reduced Staffing (Limited functionality) ¥ No Staffing (Lights out operations) Potentialÿ Savings as % of DAAC Operations Support (Through Oct 2002) 24 Hr Shift Positions at DAACs Human Interface vs Automation Cost/Performance Roadmap Analysis of Requirements for ÒTall PoleÓ Products Analyzing latest product survey that reflects increase in processing and storage/distribution requirements ¥ Currently evaluating 6 Òtall poleÓ products from Processing perspective (2 MODIS, 4 MISR) - Determining impacts for various processing architecturesÿ - Also evaluating impacts of storage requirement increases ¥ Working with science teams - Evaluating and understanding estimation processes - Evaluating applicability of parallel processing architectures for algorithms - Working on plans for joint evaluation of algorithms prototypes on different processing architectures (Networked Workstation prototype using Parallel Virtual Machine (PVM) product planned for STL) ¥ Plan future analysis of complete product survey in 1st quarter 94 Tall Pole Products Price/Performance Curves Mini-Super Price/Performance Tall Pole Processing Impact Archive & Storage Curves Tall Pole Storage & Archive Impact Conclusions of Tall Pole Analysis With expected significant decreases in price/performance for processing technology, purchase commitmentsÿ should be delayed as long as possible In near-term, expect price/performance of processing to decrease much faster than storage technology ¥ Implies periodic evaluation of process vs storage decisions ¥ Previous programs indicate need for processing capacity almost always increases over the life of the program If processing capacity of MPP architectures can be effectively utilized it may be the proper solution ¥ Expect additional developmentÿ effort ¥ Possible use of network or cluster workstation architecture may be more effective Cost/Performance Roadmap Options for Standard Product Generation Issues Affecting Process vs Storage Product Requirements Processing Load vs. 5-year Storage Volume Cost Leverage (Single Product) Current Views and Recommendations for Process vs Storage Future Plans Complete analysis of new product survey in 1st quarter 94 Develop system model used to perform detailed sensitivity analysis ¥ Required processing, caching, archive and communication bandwidth ¥ Determine expected system response for various service requests ¥ Determine impacts of user/data modeling input ¥ Perform more complete analysis of process vs storage trade Plan to have initial system model developed during first 6 months of 94 to support system architecture/design analysis