Bachelorarbeit BCLR-2018-71

Günthör, Johannes: Detection of unintended configuration changes in continuous deployment pipelines.
Universität Stuttgart, Fakultät Informatik, Elektrotechnik und Informationstechnik, Bachelorarbeit Nr. 71 (2018).
123 Seiten, englisch.

Once an Amazon Web Services employee took numerous servers offline that should have stayed online. The resulting US-East outage originated from a single integer that was incorrectly inserted in a configuration file. The configuration file defined the number of servers that should be up and running. The change was legal, but the number of servers that where taken offline was to high. We took the idea of such simple errors and applied them to continuous deployment pipelines. Continuous deployment pipelines are the next evolutionary step in continuous build pipelines. The Amazon Cloud is a highly automated environment. Pipelines are similar to the Amazon cloud a highly automated environment. Small errors can have potential catastrophic outcomes. We interviewed two experts in two differently sized companies which are using continuous integration and continuous delivery pipelines. Based on the information that was provided by both experts about the state of current continuous pipelines, we derived influence factors that are problematic in continuous deployment pipelines. The discovered influence factors already exist in currently used continuous integration/delivery pipelines were they do pose less significant threats than in continuous deployment pipelines. The enhanced automated deployment process of continuous deployment pipelines are making these factors problematic. We developed classification and improvement methods for each of the discovered influence factors. These methods can be used to strengthen a pipeline against unintended configuration changes.

Volltext und
andere Links
Abteilung(en)Universität Stuttgart, Institut für Softwaretechnologie, Sichere und Zuverlässige Softwaresysteme
Betreuervan Hoorn, Dr. André; Düllmann, Thomas; Endres, Christian
Eingabedatum16. Mai 2019
   Publ. Informatik