Lean Six Sigma and Software Development Process Report

Introduction

Six Sigma (6 σ) is a process output quality measurement practice. The benchmark for Six Sigma measurement is 3.4 defects per one million opportunities (DPMO). “An opportunity is defined as a chance for non-conformance, or not meeting the required specifications” (Ferrin). To enhance the process output quality the Six Sigma metric is used in defect reduction effort. Sigma is a statistical term that measures deviation from perfection (Ferrin). Lean is a process of uncovering the defects with special attention to reduce waste. This is achieved by reducing the complexity of SDLC processes with small iterations, well-defined requirements and incremental development (Hallowell).

Six Sigma Methods

Six Sigma methods used for defect reduction in the SDLC process output are:

DMAIC

This Six Sigma method is applied to improve existing software. The passage through tollgates between different phases of this method is reviewed by higher Six Sigma authority (Gack).

• Define – Define CTQ attributes and convert them to actionable goals, draw the project charter, draw the process map with inputs and outputs for every SDLC process involved for each CTQ attribute .
• Measure – For every CTQ attribute collect the required data, set the software quality performance improvement goals and calculate the defect variance introduced due to non-conformance of this attribute.
• Analyze – Refine the software capability in order to include quality improvements, the goal set may be included in the current project scope or may require division into multiple short duration goals. The measured data must be segmented to identify patterns that will determine factors (x) for every goal, e.g. if all goals slip the deadline then scheduling or lack of expertise in the team may be identified as factors effecting the quality. The data patterns may suggest that a design factor (x) is to be investigated for quality enhancement. The defect opportunities that have resulted in quality outliers may be treated as special cause. The financial cost of the opportunity is calculated to determine the effect of quality variance on the cost, e.g. Software capability or performance may be degraded due to the defect.
• Improve – After factors (x) have been identified the alternative solutions for defect opportunities are modeled & simulated to find the best solution. The best solution for optimized factors is implemented.
• Control – A control plan to monitor the relationship between factors (x) and CTQ attributes (Y) is developed and executed. The improved software quality statistics are collected and compared with the benchmarks set in define & measure phase. The software must be validated to confirm that no new vulnerable opportunities are created as a side-effect of this improvement. After tollgate review the developed software may be released and development records are archived.

DMADV

This Six Sigma method is applied for new software development.

• Define – Define the goals for the new software development.
• Measure – Identify attributes critical for the customer (CTQs), the software capability requirements, the software acceptance benchmark set by the customer and processes for software stability.
• Analyze – In every SDLC process during software development analyze all alternatives to find the best solution for the defined goals & measurements.
• Design – In all design activities optimize the design and verify the design by applying usability methods.
• Verify – Before the developed software is released, extensive testing is done to verify that benchmark measurements are met.

Six Sigma Advantages

The use of Lean Six Sigma methods and tools in software development increases employee efficiency, reduces defects and process repetitions thus enforcing control on deadlines and money expenditure. Long projects are converted into multiple short duration projects for DMAIC or DMADV methods, this provides

• Control on performance quality by counting number of defects and variance from benchmarks, e.g. Number of database access failure per hour due to server capability.
• Monitoring of defect opportunities by determining type of cause, e.g. frequent web server busy/down status observed by web users due to network congestion.
• Detailed CTQ attributes can be set for short projects thus shortening analysis phase, e.g. larger projects will provide brief requirements statements that are elaborated in the analysis phase.
• Benchmarks and variance measurements to determine defect cause factors, e.g. database access delay due to network delay and memory read/write cycles, the round-trip delay and memory access cycle times at full system capacity are used as benchmarks.
• Statistical analysis of defects & performance to avoid wasteful SDLC process iteration cycles, e.g. Architectural or design defects suggest better planning and more emphasis on these processes.
• Better management with heuristic data, tollgate reviews & realistic tools. Defects find and fix prediction for project cost & delivery estimation.

Six Sigma Disadvantages

The conversion of long projects into smaller projects may result in:

• Human resource crunch, e.g. more number of software engineers are required to develop multiple projects simultaneously.
• Higher financial cost, e.g. more hardware and human resources are required for simultaneous projects.
• Longer schedule for entire project completion, e.g. If shorter projects are executed sequentially the overall project duration may be longer as compared to one single project or multiple simultaneous short projects.
• Management overhead due to more statistical data, e.g. since more data is available better performance control and quality expectations are set by the consumer.

The employees must be trained for use of Lean Six Sigma methods and tools. Higher Six Sigma authority such as Black Belt or Master Black Belt is required for leading Six Sigma implementation effort. The Lean Six Sigma training may add to the software development cost.

Six Sigma tools

1. Benchmarking – Benchmarking defines standards for best practices to be followed in software development. The benchmarks are used to measure the quality variance in both DMAIC and DMADV methods. In the former method the measured statistics of CTQ attributes are compared with benchmarks to calculate the quality variance and define goals for improvement. In the DMADV method the developed software is verified against the set benchmarks. The benchmarks could be set for cost, schedule, functional capacity and performance of the software. The tools used for benchmark calculation may vary for CTQ attributes, e.g. Market standards, lab measurements, etc.
2. Control charts – The control charts are used in the software development to determine the common and special causes for variation to quality. In DMAIC method the control charts are used to analyze the measurements to determine the type of quality variation cause. The quality controls are then applied to the determined cause.
3. Correlation analysis – When relationship between two factors is to be established in either DMAIC or DMADV methods, correlation analysis tools are used. These tools are used to determine the variance in CTQ attributes to establish positive or negative relationship between the two factors. The tools may be used for both management and functional factors of the software. E.g. the former is dollar cost vs. time, the latter is software capability vs. performance due to memory or processing requirements.
4. Flowcharts – The flowcharts are used in the SDLC processes analysis phase such as requirement definition and design. The flow charts provide a graphical representation of the activity, inputs and outputs. Flowcharts are used in the analysis, improve/design phase of both DMAIC and DMADV methods for process optimization and as instruction manuals.
5. Modeling and Simulation – This tool is used in both DMAIC and DMADV methods improve and design phases respectively to validate the function and performance of prototype/developed software. This tool is used to measure the functional capacity and performance of the software for benchmarking and quality control.
6. Scatter diagram – In DMAIC method the correlation between CTQ attribute factors is calculated to analyze the impact of these factors on the quality improvement. The correlation r value is calculated by the scatter diagram drawn with the measured statistics, the perfect 1 value suggests a linear relationship and a low value suggests no relationship (Sloan). The r value is used to determine the factors that must be improved or controlled for better quality. In DMADV method the scatter diagrams are used to determine the correlation between requirements and design attributes to select the best solution.

Works Cited

1. Control Charts. Six Sigma. Web.
2. Ferrin, M. David., David Muthler. & Martin J. Miller. Six Sigma and Simulation, so what’s the correlation? ACM 2002.
3. Frye, Colleen. Six Sigma helps reduce software defects, improve quality.
4. Gack, Gary. Six Sigma Software Development Case Study. isixSigma. Web.
5. Hallowell, L. David. Software Development Convergence: Six Sigma-Lean-Agile. isixSigma. Web.
6. Heinz, Lauren. Using Six Sigma in Software Development. SEI. 2004. Web.
7. Simon, Kerri. DMAIC versus DMADV. isixSigma. Web.
8. Sloan, Daniel. Scatter Diagrams and Correlation Analysis. isixSigma.