On the Homogeneous Multiprocessor Virtual Machine Partitioning Problem - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2013

On the Homogeneous Multiprocessor Virtual Machine Partitioning Problem

Stefan Groesbrink
  • Function : Author
  • PersonId : 1001388


This work addresses the partitioning of virtual machines with real-time requirements onto a multi-core platform. The partitioning is usually done manually through interactions between subsystem vendors and system designers. Such a proceeding is expensive, does not guarantee to find the best solution, and does not scale with regard to the upcoming higher complexity in terms of an increasing number of both virtual machines and processor cores. The partitioning problem is defined in a formal manner by the abstraction of computation time demand of virtual machines and computation time supply of a shared processor. The application of a branch-and-bound partitioning algorithm is proposed. Combined with a generation of a feasible schedule for the virtual machines mapped to a processor, it is guaranteed that the demand of a virtual machine is satisfied, even if independently developed virtual machines share a processor. The partitioning algorithm offers two optimization goals, required number of processors and the introduced optimization metric criticality distribution, a first step towards a partitioning that considers multiple criticality levels. The different outcomes of the two approaches are illustrated exemplarily.
Fichier principal
Vignette du fichier
978-3-642-38853-8_21_Chapter.pdf (298.22 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-01466677 , version 1 (13-02-2017)





Stefan Groesbrink. On the Homogeneous Multiprocessor Virtual Machine Partitioning Problem. 4th International Embedded Systems Symposium (IESS), Jun 2013, Paderborn, Germany. pp.228-237, ⟨10.1007/978-3-642-38853-8_21⟩. ⟨hal-01466677⟩
54 View
66 Download



Gmail Facebook X LinkedIn More