If you are in the field of networking, chances are you have dealt with QoS (Quality of Service) countless times and you understand what that entails — at least at the fundamentals level. But, what the hack is dynamic QoS?
In this article we will expand on the capabilities and uses of QoS including some recent variations of QoS that have been popularized recently by SD-WAN. Then we will dive into dynamic QoS, which can be best described as the cutting-edge implementation of providing service guarantees for network services.
If you envision QoS as a set of various knobs on your networking gear, say your firewall or router, then dynamic QoS is the ability for those knobs to have auto-drive mode. In other words, those QoS knobs can be adjusted without necessarily requiring human intervention. To better understand the value and benefits of dynamic QoS, it makes sense to talk about the specific QoS knobs first.
As a side note, each vendor will have differing implementations of their QoS features and capabilities and sometimes may have different naming conventions, such as QoE (Quality of Experience) or QoX or something similar, which are slight variations on the definition of QoS. But for the purposes of this article we will stick with the more common term QoS.
QoS can be viewed as a traffic cop (QoS manager) who polices the cars flowing (IP traffic flows) through the roads (WAN connections). Similarly, the various QoS rules can be viewed as the various rules that affect the flow of traffic, such as the ramp meter letting cars enter a highway at a certain pace, the speed limit on the highway and various other traffic rules that affect the travel time and reliability of the vehicles (IP packets). In today’s networks, QoS rules such as traffic shaping and bandwidth reservations are implemented statically whereby the network admin can enter the traffic rules governing that network. Continuing with our analogy, for example you may enter a rule of 70 miles per hour speed limit on your WAN links for the traffic or create reserved car-pool lanes.
For a recap of how to choose your QoS router, read this post.
So how is dynamic QoS any different?
Imagine now that you are not just entering the QoS rules into your network device, but by switching on an option, you are letting your network auto-adjust the traffic rules as it sees appropriate. For example, if there is no congestion in the WAN link that your IP packets are traveling in, the speed limit can be raised from 70 miles per hour, to say 100 miles per hour or higher. In a similar fashion, whenever your WAN highway is not being fully saturated, the dynamic QoS can decide to change the pace of cars that are let into the WAN highway. Are you still with me regarding our car traffic lanes and WAN networks analogy? Good.
Dynamic QoS is a game changer for SD-WAN
Once the dynamic QoS capabilities are available within an SD-WAN solution, it completely changes how networks are monitored and managed. It changes from reactive to proactive, from manual to automated and as a natural consequence it changes your network from costly to cost-effective, both in terms of deployment and operations. A crippling network brown-out in your business network, with the help of dynamic QoS, can now be converted into a non-emergency maintenance ticket. Better yet, if the brown-out can be fully mitigated, all the associated cost of this network emergency can be 100% avoided. Continued access to your cloud services means zero downtime and happy end-users as well as happy managers.
Ok, sounds good but how does dynamic QoS work under the hood?
QoS relies on measurements of various network metrics such as packet loss, latency, jitter, throughput rate and so on to make sense of the current network environment and comparing that to the target. As an example, in a latency optimizing QoS setup, if a WAN link has a spike in one-way latency, the system should realize that the latency metrics for the QoS are no longer met and should take action to correct this situation (the specific action taken by dynamic QoS is outside the scope of this article, however, one can envision actions such as traffic shaping to limit cross traffic, or intelligently steering traffic to a better route). So, similar corrective and dynamic actions will be taken by the dynamic QoS module without requiring any human intervention for the corrective measure. That is really the core dynamism of the dynamic QoS, i.e. adaptively and intelligently being able to react and resolve network problems.
Limitations of dynamic QoS
Yes, admittedly, dynamic QoS sounds too good to be true. Is it really that futuristic and bulletproof? The short answer is — not quite yet. Dynamic QoS in SD-WAN is leaps and bounds better than static QoS in legacy routers and firewalls, however, there is still upcoming technologies in the pipeline that will take networking to yet another level. The vast amount of information and measurements in the network today are processed primarily for anomaly detection. While this is a great and direct use of ML (Machine Learning) and AI (Artificial Intelligence) in networking, it is just scratching the surface with those technologies. ML/AI will be able to not only detect anomalies in networks but also react in an optimum way, perhaps one day in a way no human can.
Cahit Akin, CEO, Mushroom Networks, Inc.
Mushroom Networks is the provider of Broadband Bonding appliances that put your networks on auto-pilot. Application flows are intelligently routed around network problems such as latency, jitter and packet loss. Network problems are solved even before you can notice.