Network functions virtualization (NFV) holds the potential to reduce capex and opex, accelerate time-to-market, and deliver agility and flexibility. But if the transition from physical to virtual appliances is not done correctly, virtual appliance performance may be worse than their physical counterparts’. A new white paper from Ixia looks at how virtual network function (VNF) performance testing can help determine best practices for a smooth and profitable migration.
“How To Compare Virtual Devices (NFV) vs. Hardware Devices: Testing VNF Performance” is part of Ixia’s NFV thought leadership series, which aims to teach people how to transform their network architecture through testing and optimization rather than installing new equipment.
The basic concept of NFV proposes that network functions like routing, firewalls, load balancing, and DPI run on standard servers instead of expensive, proprietary appliances. The white paper says NFV architectures often face significant bottlenecks through vSwitches and the VNFs themselves that can constrain performance and minimize NFV benefits.
Ixia takes a look at each bottleneck and outlines how VNF testing can help networks overcome them. The paper discusses two main vehicles for testing: 1) the BreakingPoint Resiliency Score, which establishes standards against which network performance and security (physical or virtual) can be measured; and 2) the Data Center Resiliency Score, which helps determine the optimal resource allocation required to support a desired number of users.
The white paper highlights practical applications for each testing vehicle. For example, the Resiliency Score allows organizations to benchmark performance, security, and scale of VNFs and compare them to their physical counterparts. In the white paper sample case, a bake-off pits a traditional firewall against its virtual counterpart and discovers that the default configuration of the software-based firewall leads to impaired performance compared to the hardware-based model.
The white paper explains how Ixia then changed one parameter at a time (as opposed to changing complete system configurations), retesting after each change, to isolate bottlenecks in the VNF performance. It was only after fine-tuning the virtual environment that the virtual firewall delivered the performance expected.
As paradigm shifts become the norm in data communications, the white paper argues, such performance testing becomes increasingly important to maximize the benefits of NFV throughout the deployment lifecycle. To learn more about VNF testing and how to optimize NFV performance, request your copy of “How To Compare Virtual Devices (NFV) vs. Hardware Devices: Testing VNF Performance.”