Most vendors claim that their SD-WAN solutions can deliver Quality of Service (QoS), but is this really true? The concept of QoS has been much discussed in recent years and it really refers to objective attempts to measure network service-level parameters, such as packet loss and jitter, and control the impact of these characteristics on performance.
Of course, QoS can mean different things, but if we consider one useful definition from Wikipedia - “Quality of service is the ability to provide different priority to different applications, users, or data flows, or to guarantee a certain level of performance to a data flow”, then we would argue that most SD-WAN solutions don't really deliver true Quality of Service at all.
And, there's another dimension we need to consider — Quality of Experience (QoE), which, happily, does have a universally agreed definition, under ITU-T Recommendation P.10. According to this, QoE relates to the perception users have regarding the services they use that are delivered across a network.
This is important, because it affects what users are actually doing — QoS may be obtained for the network as a whole, but people are usually trying to do different things across that network, so the needs of the business applications they are using must also be considered. Some applications are more important than others or have more stringent performance requirements, so we must consider if users can effectively undertake the tasks they are supposed to, particularly when the applications are based in the cloud.
In this short series of articles, we'll explore this further and show why today's SD-WAN solutions don't necessarily deliver QoS and may not meet QoE expectations either. We'll also see how QoS and QoE can be realized in practice.
Quality of Service (QoS) Myths
Many SD-WAN solutions claim that three key techniques are required to deliver QoS:
- Forward Error Correction (FEC);
- Path selection; and
These can deliver improved performance, but is it really QoS?
Performance Myth #1 — FEC
First, let's look at FEC, which is widely adopted. With FEC, duplicate packets are sent to ensure successful application delivery and improved performance. It's typically used with real-time traffic, in case some packets aren't delivered. As a result, FEC actually increases the bandwidth needed and increases costs.
That can lead to inefficiencies and, while it may mean that there's less packet loss, it doesn't necessarily result in enhanced QoS. FEC doesn't consider the competing needs of different applications, so by duplicating packets, no allowance is made for which packets are really important. This matters when we consider user QoE. As we said, one application may be more important than another, which means increasing traffic load for all packets may negatively impact a critical business service by overloading the network or compromising delivery of key packets from more valuable services.
So, if not all packets are equal, a policy of simple duplication may not be the best approach. Instead, it may be better to optimize delivery from the outset, not by providing duplication. For example, InfoVista's Ipanema solutions handle traffic more efficiently on a session by session basis, rather than across applications globally, without the need to send duplicate packets.
With InfoVista, applications are delivered according to their business priority. This means that network resources are used in the most efficient way possible from the outset, eliminating the need to send duplicate packets and conserving bandwidth resources. Unlike other vendors that offer FEC-based techniques, InfoVista reduces traffic volume. This approach helps support true QoS, while also paying attention to the QoE outcome desired!
Stay tuned for more…
Read Part 2 to learn about the myth around Path selection! Does path selection contribute to QoS?
Looking for more SD-WAN myths to bust? Watch on-demand webinar: Top 7 SD-WAN myths — busted! where Zeus Kerravala, founder of ZK Research, joined us for to expose the truth about top SD-WAN myths!