Saturday, July 14, 2012

Agile Process Performance Model supporting CMMI- Product Quality Assessment

Image Courtesy- http://www.freedigitalphotos.net

Continuous Integration System can be used for effectively assessing the product quality in a very systematic manner. Each product team can select some of the key product quality parameters that are decisive about the quality. In this post we would propose a simple way of assessment which can be built up further by teams to come up with a kind of performance model based on the data that is captured.

Aspect Details Measurement/ Formula Expected Value Actual Value
Static Code Errors and Warnings Comprises of selected set of rules OR All rulesTotal no. of errors and warnings open 0 << From CI System>>
Function Coverage Can be measured for set of Functions OR for all FunctionsNo of Functions Covered > 90% << From CI System>>
Feature Stability MeasureList of important features OR All features. Feature Stability Average of the Daily Feature pass rate100%<< From CI System>>

Above aspects are just suggestions. There can be other parameters like Code Complexity, Maintainability Index, Line Coverage, Branch Coverage and others. Each team can customize the product quality parameters and measure it accordingly. Each build can be monitored on a daily basis for the key product quality parameters and the historical data can be stored. Using statistical or analytical techniques process performance model can be built based on the dependent, independent and controlled variables among these parameters. Based on the process performance model the product quality can be predicted / assessed at a sprint or release level.

Reference:
http://ppqc.blogspot.in/2008/08/help-understanding-process-performance.html

Friday, July 13, 2012

Agile Process Performance Model supporting CMMI- Feature Stability Prediction

Image Courtesy- http://www.freedigitalphotos.net
The intention of a process performance model is to identify a quantitative understanding of the performance of  the sub-processes which in turn  helps in providing process performance data, baselines and models to quantitatively manage the organization's projects. Process Performance Models are used to represent past and current process performance and to predict future results of the process. At the first glance this looks like very data intensive and very difficult for the agile projects to maintain. In this post we would give some insight into how continuous integration system can support in building feature stability prediction model.


Feature Stability Assessment is one method / technique based on which the feature stability can be assessed and there can be some prediction done for the quality of the product.

When the test cases are being designed, the following naming convention can be used:
<<FeatureName>>_<<UT_TestSuiteName>>
Each Test Suite can be further mapped to the test cases level.
In each build the suite and the test cases which are failing can be monitored closely. Also there can be a trend which can be brought out about the features which are defective in last few builds.

Example-

Feature Stability Assessment can be done at Build Level and at the release level to check the quality of the features being built over a period. This can help in better prediction of the quality of the features and decisions can be taken to focus on specific improvements at feature level. Based on these factors regression models can be built to predict the quality of a product effectively.

Thursday, July 12, 2012

CMMI and Agile- Need a balanced approach for organizational success

In our association with various teams of different sizes and different domains, we have come across many questions related to adoption of CMMI® and Agile. Questions like, "We are CMMI® Level 3 Organization, how can we be agile", " Our company is doing Agile, but our customers are looking for CMMI® Level 5, how do we adopt both" etc. In this post and the subsequent posts in this category, we would provide our perspective about CMMI and Agile and how both the good practices can be merged to bring organizational success.
Image Courtesy- http://www.freedigitalphotos.net
Any model or methodology is a set of good practices that have been proven and can be adopted for better results in similar kind of work. All companies wants to deliver the products and services better, faster and cheaper. Nearly all organizations have found that the products and services have varied levels of complexity. It is unusual today for a single organization to develop all the components that compose a  complex product or service. More commonly, some components are built in-house and some are acquired. The integration of these components needs better planning and strong engineering practices.Organization must be able to manage and deliver such complex development and maintenance.

CMMI®-Dev comprises of 22 process areas. These process areas covers aspects related to Engineering, Planning, Support and Organizational processes.  Agile is a way of thinking about software development. The canonical description of this way of thinking is the Agile Manifesto, a collection of 4 Values and 12 principles. (http://www.thinkinginagile.com/2011/03/agile-manifesto.html). Agile Methods are available in different flavors like SCRUM, XP, Lean Software Development (Kanban), DSDM, Crystal and others. Most of the teams adopt SCRUM and XP together. SCRUM comprises of the management practices and XP consists of the engineering practices.

One important fundamental which needs to be noted is : A method or process is a way of working. Whenever we do something, we are following a process. Some processes are written, as-when we are assembling a machine but others are adhoc and informal, as-when sometimes we clean our house. Most of members we have worked have said that CMMI is about excessive documentation. In fact this is not correct about CMMI. As per Appraisal Requirements for CMMI®Version 1.3 (ARC, V1.3) Section 4.3.4," the finding can be derived from objective evidence seen or heard during data collection sessions." This means that for the rating purpose when the data is being collected Objective evidence is seen or heard or observed during the appraisal time. While following the Agile practices there are many instances where the documentation is less. Some examples are given below :
  • Daily Standup Meeting is done where the actions are being tracked on a daily basis unlike Weekly Meeting.
  • Simple Design is achieved through XP practices like Test Driven Development, Refactoring unlike detailed low level design documents.
  • Continuous Integration is practiced instead of dedicated Integration testing phase.
Above list may go on and on. In fact these practices are less documentation intensive but it can ensure that the management and engineering level practices are followed well. It is really important to understand that CMMI looks for the intent of the process area rather than documentation. So CMMI and Agile compliment to each other. The basic issue is about the right understanding about both the set of good practices.

CMMI® provides set of process areas that covers all the activities of software development. Agile methods like SCRUM and XP provides simplified solutions for management and engineering practices. Collection of agile best practices through these methods makes the delivery process fast and robust. It is important to adopt both CMMI and Agile practices together rather than bringing conflicts with each other. All it requires is to balance and customize the process based on the business and the team working for the development. In the next posts we would provide insight about how the good practices of  agile and the CMMI® process areas can be integrated effectively for project and organization success.