RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación
Artículo científico (JournalArticle)
Automatic memory-based vertical elasticity and oversubscription on cloud platforms
Hypervisors and Operating Systems support vertical elasticity techniques such as memory ballooning
to dynamically assign the memory of Virtual Machines (VMs). However, current Cloud Management
Platforms (CMPs), such as OpenNebula or OpenStack, do not currently support dynamic vertical elasticity.
This paper describes a system that integrates with the CMP to provide automatic vertical elasticity to
adapt the memory size of the VMs to their current memory consumption, featuring live migration to
prevent overload scenarios, without downtime for the VMs. This enables an enhanced VM-per-host
consolidation ratio while maintaining the Quality of Service for VMs, since their memory is dynamically
increased as necessary. The feasibility of the development is assessed via two case studies based on
OpenNebula featuring (i) horizontal and vertical elastic virtual clusters on a production Grid infrastructure
and (ii) elastic multi-tenant VMs that run Docker containers coupled with live migration techniques.
The results show that memory oversubscription can be integrated on CMPs to deliver automatic
memory management without severely impacting the performance of the VMs. This results in a memory
management framework for on-premises Clouds that features live migration to safely enable transient
oversubscription of physical resources in a CMP.
© 2015 Elsevier B.V. All rights reserved., The authors would like to thank the Spanish "Ministerio de Economia y Competitividad" for the project CLUVIEM (TIN2013-44390-R) and the European Commission for the project INDIGO-DataCloud with grant number 653549.