Bearbeiter | (nur für Mitarbeiter:innen einsehbar) |
Betreuer | Dr. Stefan Brenner |
Professor | Prof. Dr. Rüdiger Kapitza |
Projekt | |
IBR Gruppe | DS (Prof. Kapitza) |
Art | Masterarbeit |
Status | abgeschlossen |
Infrastructure clouds are used by a wide variety of users in the scientific and commercial field. While high availability for applications running inside the cloud can be achieved by the replication of virtual machines, the underlying infrastructure itself is still vulnerable to failures. This thesis aims at improving the high availability features of Openstack by providing means to secure selected core services against failures. These services include the central messaging bus (RabbitMQ), the cloud controller as well as it's database and the respective public API for user interaction. In a first step, possible failure scenarios within the architecture have to be located and evaluated according to their severity. The effects of faults of other cloud components will be classified into different categories (e.g. infrastructure partially affected or completely unusable). The next step comprises of developing fault-tolerant versions of the core services reusing standard technologies of other Openstack projects like "Heat" if possible. The individual software components will be evaluated with respect to extensibility and reuseability. Possible approaches typically include active or passive replication on two or more machines with automatic failover procedures. Third, a simulation environment will be installed, in order to demonstrate the fault-tolerance of the modified services. The developed strategies will then be compared to approaches used by other IaaS projects like Eucalyptus, Nimbus or OpenNebula, outlining the respective advantages and disdvantages. |
Technische Universität Braunschweig
Universitätsplatz 2
38106 Braunschweig
Postfach: 38092 Braunschweig
Telefon: +49 (0) 531 391-0