Testing SD-WAN To Validate Application Performance And Determine Best Solutions


SD-WAN is one of the latest technologies to originate from the new wave cloud of software defined whatever. Traditional wide area networks (WANs) were not developed or architected for the cloud, so as cloud adoption grows evermore popular at the enterprise level, so does the need for cloud-based WANs. Boasting the ability to optimise the capacity, flexibility and reliability of any managed network, and to reduce the number of network connections and devices within an enterprise, SD-WAN may even make WAN technologies like Multiprotocol Label Switching (MPLS) obsolete.

The SD-WAN market is expected to be valued at more than $8 billion USD in 2021 (IDC). Yet enterprises are hesitant to adopt over concerns such as potentially degraded performance, deployment costs, and security.

Most readers have probably been exposed to enough of the subject to get a basic idea for how SD-WAN works. For those who need a refresher: a controlling device, or SD-WAN controller, continuously measures network conditions like packet loss, latency, jitter, and the available bandwidth of each path it is set to monitor. It then routes traffic across the best available path to ensure an optimal connection even if WAN services experience impairment.

SD-WAN vendors use their own proprietary algorithms to determine the method and means of data transfer. For example, Silver Peak’s Unity EdgeConnect creates application overlays comprising multiple WAN transports to deliver higher application availability, and Riverbed’s SteelConnect uses DPI-based application awareness to distinguish between business-critical and recreational traffic. Choosing the most appropriate solution for your enterprise can be difficult, especially when considering the networking requirements of your business, the size of your budget, and the distances data is being transferred, among other concerns.

SD-WAN vendors usually test and validate their solutions throughout the development cycle to identify bugs, performance issues and generally, how the product will perform under real world conditions – before they deliver a product to market.

However, vendor development and pre-deployment testing can never fully test every possible scenario each enterprise faces. Enterprises considering moving to SD-WAN need to test and validate each vendor solution to be considered based on all possible network and application scenarios that their stakeholders will face.

A purpose-built WAN emulation appliance can be a tool well worth the investment, considering the high costs of network downtime and stakeholder dissatisfaction. These devices can replicate a live network in a lab environment so that application performance can be reliably measured, validated, and optimised before going live. Testers can also apply impairments, such as delay, loss, and jitter, to the emulated network in order to observe packet flow through the WAN and how the SD-WAN system reacts to such conditions. Once an investment is made, WAN emulators can validate change management and replicate end-user issues that ultimately arise.

Testing can be conducted quickly, accurately and locally with a network emulator, and validating application performance between multiple sites doesn’t require travel time or several days of planning and coordination. Everything is done within the lab, without a router or modem. Plug in the client, server, and management devices to their own respective ports on the emulator, log-in to the web-based GUI, and configure traffic.

Live network conditions are profiled and automatically reproduced to reflect the dynamic conditions of your network, and IT departments can create an infinite amount of if/then scenarios and measure how the SD-WAN system reacts to network outages, brownouts, congestion, excessive packet loss, and other impairments that could make for a poorly performing network.

The next step is at system level, to see if the SD-WAN meets the overall network design goals. For instance, some businesses may favour SD-WAN for its failover capabilities, which can save a network if a link is ever completely down, while others prefer SD-WAN for simplified infrastructure, flexible traffic management or business continuity.

Since most network outages are characterised by high levels of packet drops and bursts of small pauses, deciding when to switch from one link to another, as well as when to switch back, can be a tough call. Different solutions’ algorithms handle these events quite differently. To get a good idea for how each system would handle a failover scenario, testing conditions should not only recreate these instances of failover, but also emulate a wide range of packet loss rates.

If your network handles multiple sites, simulate deployment of those sites within the lab to observe the limitations an SD-WAN controller may present. If you already have an MPLS network in place, manipulate its settings to alter the traffic flow and get an idea for how your network will perform without it. Conduct a series of actions users may often engage in, add and remove users, reconfigure nodes, and test everything you can.

A network is a living thing: its needs, as well as the needs of its end users, are unique. Regardless of the SD-WAN offerings being tested, the key is to determine which solution best accommodates the requirements of your company’s network and budget. SD-WAN systems manage the networks upon which the success of your business relies, so test them properly and thoroughly. A WAN emulation appliance provides the functionality and accuracy you need to accurately test and optimise these solutions to ensure that your transition will be as unfailing and pain-free as possible.

Neal Roche

Neal Roche, CEO of Apposite Technologies, has over 15 years of experience with network testing and validation companies, including management roles at Ixia and Spirent.