Modelling Failures Occurrences of Open Source Software with Reliability Growth - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2010

Modelling Failures Occurrences of Open Source Software with Reliability Growth

Résumé

Open Source Software (OSS) products are widely used although a general consensus on their quality is far to be reached. Providing results on OSS reliability - as quality indicator - contributes to shed some light on this issue and allows organizations to make informed decisions in adopting OSS products or in releasing their own OSS. In this paper, we use a classical technique of Software Reliability Growth to model failures occurrences across versions. We have collected data from the bug tracking systems of three OSS products, Mozilla Firefox, OpenSuse and OpenOffice.org. Our analysis aims at determining and discussing patterns of failure occurrences in the three OSS products to be used to predict reliability behaviour of future releases. Our findings indicate that in the three cases, failures occurrences follow a predetermined pattern, which shows: a) an initial stage in which the community learns the new version b) after this first period a rapid increase of the failure detection rate until c) very few failures are left and the discovery of a new failure discovery is rare. This is the stage in which the version can be considered reliable.
Fichier principal
Vignette du fichier
rossi-etal.pdf (432.54 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01056050 , version 1 (14-08-2014)

Licence

Paternité

Identifiants

Citer

Bruno Rossi, Barbara Russo, Giancarlo Succi. Modelling Failures Occurrences of Open Source Software with Reliability Growth. 6th International IFIP WG 2.13 Conference on Open Source Systems,(OSS), May 2010, Notre Dame, United States. pp.268-280, ⟨10.1007/978-3-642-13244-5_21⟩. ⟨hal-01056050⟩
92 Consultations
263 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More