TY - GEN
T1 - Hvbench
T2 - 2nd IEEE International Conference on Network Softwarization, NetSoft 2016
AU - Sieber, Christian
AU - Blenk, Andreas
AU - Basta, Arsany
AU - Kellerer, Wolfgang
N1 - Publisher Copyright:
© 2016 IEEE.
PY - 2016/6/30
Y1 - 2016/6/30
N2 - Software-defined networking (SDN) introduces a split between the forwarding plane and the control plane of modern network equipment. Furthermore, OpenFlow, as an open interface for SDN, fosters the development of logically centralized network operation systems (NOS). To allow multiple network operation systems accessing the same network, there is the concept of SDN hypervisors. Little is known about the performance characteristics of SDN hypervisors. Furthermore, there is a lack of scalable and realistic hypervisor benchmark tools capable of emulating dynamic load scenarios. In this paper, we present an extensible and distributed SDN hypervisor benchmarking framework based on flexible statistical request generators. The framework can be scaled out horizontally to multiple compute nodes that are centrally controlled and reconfigured at runtime. We present preliminary measurements of the CPU resource consumption of a hypervisor in a virtual environment. The results show that the performance characteristics of the hypervisor are different for dynamic load scenarios compared to static benchmarks. Furthermore, the results show that for the same overall request rate, multiple NOS increase the CPU load considerable compared to a single NOS.
AB - Software-defined networking (SDN) introduces a split between the forwarding plane and the control plane of modern network equipment. Furthermore, OpenFlow, as an open interface for SDN, fosters the development of logically centralized network operation systems (NOS). To allow multiple network operation systems accessing the same network, there is the concept of SDN hypervisors. Little is known about the performance characteristics of SDN hypervisors. Furthermore, there is a lack of scalable and realistic hypervisor benchmark tools capable of emulating dynamic load scenarios. In this paper, we present an extensible and distributed SDN hypervisor benchmarking framework based on flexible statistical request generators. The framework can be scaled out horizontally to multiple compute nodes that are centrally controlled and reconfigured at runtime. We present preliminary measurements of the CPU resource consumption of a hypervisor in a virtual environment. The results show that the performance characteristics of the hypervisor are different for dynamic load scenarios compared to static benchmarks. Furthermore, the results show that for the same overall request rate, multiple NOS increase the CPU load considerable compared to a single NOS.
UR - http://www.scopus.com/inward/record.url?scp=84979997833&partnerID=8YFLogxK
U2 - 10.1109/NETSOFT.2016.7502475
DO - 10.1109/NETSOFT.2016.7502475
M3 - Conference contribution
AN - SCOPUS:84979997833
T3 - IEEE NETSOFT 2016 - 2016 IEEE NetSoft Conference and Workshops: Software-Defined Infrastructure for Networks, Clouds, IoT and Services
SP - 403
EP - 406
BT - IEEE NETSOFT 2016 - 2016 IEEE NetSoft Conference and Workshops
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 6 June 2016 through 10 June 2016
ER -