Automated, Cloud-based and Real Time Software Reliability Growth Modeling

Next Session: TBD

Importance and Timeliness

The industry is evolving towards a convergence of ICT and Telecommunications in which most communications software will run on generic servers the cloud. This trend is making it a requirement for software vendors to ensure continuous product integration and delivery. In order to guarantee the required levels of software reliability within such agile processes, software reliability growth modelling (SRGM) tools must also evolve. Specifically, in practice, the defect trend changes during the test phase. This is mainly due to availability of new features or functions being developed. This makes traditional single-curve SRGM unable to accurately describe defect trends. Moreover, due to the shorter development cycles, SRGM should be adaptable to shorter timescales such as using daily data rather than the typical weekly data. In both these cases, it is no longer feasible to rely on tools which require significant manual tuning. In other words, automation in SRGM, especially for relatively big projects, is indispensable.

Workshop Summary

This workshop will discuss practical aspects of software reliability measurement, prediction, and application. We will first address basic background and the state-of-the-art of SRGM. We will then introduce a recently developed cloud-based SRGM tool which, for a given project, automatically generates multiple piece-wise curves so as more accurately capture changing defect trends. The tool is applicable to different software development projects without a need for additional manual tuning. Moreover, we will also introduce a new innovative method for predicting defect arrival curve based on the new code arrival curve during the planning phase. This method is especially useful for project management to determine staff requirement for development and test. The result is a robust, consistent, flexible, fast, statistically sound approach for defect prediction for any defect data sets without human intervention. The cloud-based SRGM tool is provided as-a-service to software development teams spread across geography and time with automated, real-time, reliable and actionable insights regarding the development process. This method has been successfully applied to development projects with various size and network functional elements using traditional and Agile software development methodologies within Nokia. A large number of use cases will be discussed for effective use of the tool to assure high reliability of a software product.

Length: Two Days

Speakers: Kazu Okumoto | Rashid Mijumbi | Abhaya Asthana

Target Audience and Benefits

Practitioners, researchers and students who are interested in applying software reliability prediction to real projects. The main pre-requisite of this workshop will be a good understanding of basic software development processes as well as basic knowledge of (the need for) software reliability. Practitioners will be able to understand a software reliability prediction procedure that can be used for deciding on whether the software product is ready for delivery, and if not, the necessary amount of testing that is needed to achieve the required software quality. Researchers and/or students will be able to understand industry needs to advance future research areas in software reliability and availability.

Proposed Outline:

  • Basic SRGM measurement, prediction and application
  • State-of-the-art in SRGM
  • Automated SRGM, including demo
  • Cloud-based, real-time BRACE: Introduction and benefits, basic concept and architecture

  • BRACE Demo, including design & implementation, and use cases
  • eDPM (Early defect prediction model), including basic approach
  • eDPM case studies
  • BRACE implementation for eDPM and demo, conclusion and Q&A

Click here to register for June 14th workshop