Publications

A Case for SDN-based Network Virtualization

2021 29th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems (MASCOTS), Virtual (originally in, Houston, TX, USA), 2021, pp. 1-8, doi: 10.1109/MASCOTS53633.2021.9614291.

Gyeongsik Yang, Changyong Shin, Yeonho Yoo, Chuck Yoo

Download PDF

Abstract

Network virtualization (NV) becomes an essential technology in cloud computing that isolates network flows for tenants. However, because existing NV technologies like overlay do not enable tenants to directly program (i.e., provision, control, and monitor) network resources, software-defined networking (SDN)-based NV (SDN-NV) has been proposed. Despite its great benefits, SDN-NV has been believed to bring considerable overheads due to the network hypervisor (NH). However, to date, there is no definite performance evaluation that proves the overheads of SDN-NV. To this end, this paper comprehensively investigates the performance and overheads of SDN-NV. Our experiment results reveal that SDN-NV provides the data plane performance comparable to or even better (up to 10.5× better TCP throughput) than the existing NV technologies. Also, the results on NH show that its overheads remain mostly constant, even when the number of switches, virtual networks, or network flows increases. In short, our evaluation indicates that the overhead of SDN-NV should not deter its practical use in datacenters.

Paper summary

Measure the cost of programmable virtual networks

Network virtualization can be assessed separately for packet forwarding and for the hypervisor that manages it.

This benchmarking study investigates whether the overhead of SDN-based network virtualization prevents practical use. It separates data-plane behavior from network-hypervisor overhead so that the two costs can be examined directly.

Benchmark forwarding and hypervisors separately

The data-plane tests compare overlay networking, NAT, address mapping, tenant-ID embedding and locator embedding. Separate hypervisor tests vary switches, flows and virtual networks, using linear and fat-tree topologies.

A network hypervisor connects tenant control planes to the shared physical network.View full-size figure
A network hypervisor connects tenant control planes to the shared physical network.

Compare performance as networks grow

The tested SDN-based techniques are competitive with overlay networking in data-plane performance and exceed it in some experiments. The study also reports that hypervisor overhead mostly remains constant as its tested network dimensions increase.

The evaluation separates packet-forwarding choices from hypervisor scaling.
EvaluationCompared or varied
Data planeOverlay, NAT, address mapping, tenant-ID and locator embedding
HypervisorSwitch, flow and virtual-network counts
TopologyLinear and fat-tree networks

Key takeaways

  1. The cost of programmability has more than one source. Separating forwarding and hypervisor-processing costs makes the implementation tradeoff visible.