QoS is all about making sure that your key business applications always have the fastest path to your users on the network - never impeded, always prioritized. A feature like SD-WAN traffic shaping would seem to help, but in reality it’s too limited in its design to help your users see the performance they need.
How does SD-WAN traffic shaping work?
Fundamentally, traffic shaping is an alternative to true QoS. It doesn't prioritize key applications at all. It only focuses on limiting other applications. Which can help but can't guarantee performance for users.
Traffic shaping is designed to help real-time traffic flow unhindered at the expense of slowing down other traffic. The problem is, this is like making room for an ambulance on the highway by slowing down the other cars, but not letting the ambulance speed up and get to its destination faster once you make room for it. True QoS can do that, too, but will also increase the size of the traffic lanes for the ambulance and allow it to flow at high speed, while at the same time adjusting the number of lanes used for other cars so they still get where they need to, but just stay out of the ambulance’s way.
So if you're looking for true QoS to guarantee better user experiences, SD-WAN traffic shaping isn't going to deliver. What you're looking for is traffic prioritization. But you'll need the right approach.
Performance Misconception #3: Prioritization
Many vendors can prioritize traffic, but there are many limitations to current approaches. In most cases, the prioritization available is simply a matter of router configuration. This means that it is applied locally, at the WAN port, so it does not deliver prioritization end-to-end.
In addition, traffic is typically unpredictable and has considerable variability, but most SD-WAN prioritization solutions use the approach of static queues and so cannot handle changes in traffic volume or application mix dynamically. This can lead to inefficiencies — and not just with unexpected bursts of traffic, but also when queues are not full or when overflow occurs.
In such cases, traffic may be allocated higher or lower prioritization than is actually required, because the decision is based merely on capacity, not the nature of the traffic or the actual line conditions of the network link. Clearly, giving traffic a lower priority than is actually required can result in sub-optimal performance, while giving it too high a priority may mean that resources are taken from other traffic unnecessarily. While this may give the perception of QoS from the perspective of the network and smoothing traffic flows, it does not address the concerns of the users of the network and their objectives, which may be rather different.
A better approach …
A new approach is needed, which considers the desired outcomes (appropriate QoE for applications and users) and acts accordingly. Individual sessions need to be evaluated against the business objective of the application traffic it contains. As a result, network resources should be allocated according to real needs.
True end-to-end QoS should be independent of the network. It should allow users to control and deliver the QoE demanded of all applications in a given organization and environment. This requires several capabilities to be aligned: dynamic queues implemented above the underlying network; the ability to control the size of queues dynamically and automatically, based on network conditions; the applications; and the policy rules defined by the user. Only then can both QoS and QoE be delivered.
So, do most SD-WAN vendors offer QoS? In short, the answer is “no”. Can they improve performance? “Yes”, but enhancements are detached from the purpose of the traffic concerned and do not consider the QoE that users and applications demand.
In contrast, InfoVista takes a different approach to ensure QoS and QoE delivery. InfoVista ensures application performance, end-to-end, and offers an Application Overlay SLA, which is independent of the network. This Application Overlay is optimized for each individual application and enables performance requirements to be mapped to the business objectives of the organization, which results, not only in QoS, but the desired outcome - QoE. This is unique — and not available with existing SD-WAN solutions.
We've been doing this for years and for hundreds of large, often highly-distributed enterprise customers. Our solutions deliver session-level QoE and QoS for end-user applications, regardless of the traffic volume, application mix, source of the application delivery, or network behavior. It's the ultimate worry-free application performance solution that provides ongoing optimization as your traffic and network requirements grow and change. We'd love to show you more, so, if you want to see how to really deliver QoS and QoE, why not check out some more of our resources?
Meanwhile, this series has also raised some other interesting issues, so we'll be back soon with further QoS and SD-WAN performance insights for you. Watch out for more!
Looking for more SD-WAN misconceptions to bust? Watch our on-demand webinar, where Zeus Kerravala, founder of ZK Research, joined us to bust the Top 7 SD-WAN Myths!