Cloud Management Software that supports Scalable Resource Allocation

This thesis project focuses on the implementation of a scalable resource allocation for cloud management software. Existing cloud management software, such as Eucalyptus, OpenNebula, Nimbus, and Openstack rely on a centralized solution for managing the resources of the cloud. This approach has the well-known drawback of poor scalability, and, therefore, such software cannot be used in large-scale deployments of hundreds of thousands servers. The goal of this thesis is to decentralize the resource allocation component of an existing cloud management software in order to improve its scalability. The thesis project consists of a survey part where the student studies the architectures of existing opensource cloud-management software packages, an implementation part where the scalable solution is implemented in a selected package, and a comparison part where the performance of the extended software is evaluated against the existing one.