Performance improvements in a large scale virtualization system

Andrea Chierici, Davide Salomoni, Anna Karen Calabrese Melcarne, Alessandro Italiano, Gianni Dalla Torre
2011 Proceedings of The International Symposium on Grids and Clouds and the Open Grid Forum — PoS(ISGC 2011 & OGF 31)   unpublished
This work shows the optimizations we have been investigating and implementing at the KVM virtualization layer in our research institute based on more than a year of experience in running thousands of VMs in a production environment used by several international collaborations. These optimizations increase the adaptability of virtualization solutions to demanding applications like those run in our institute (High-Energy Physics). In particular we will show what is the best solution to share
more » ... ge space from an hypervisor to the virtual machines running on top of it, trying to avoid problems that modern storage cluster have. We will also show how different the performance can be comparing real and virtual machines, measuring access to computing, storage and network resources, trying to provide hints on server configuration where possible. This work has been driven by the project called Worker Nodes on Demand Service (WNoDeS), developed by INFN, a framework designed to offer local, grid or cloud-based access to computing and storage resources, preserving maximum compatibility with existing computing center policies and work-flows.
doi:10.22323/1.133.0049 fatcat:m3o7glxfdjenln4z4ezozmt5xe