I was fortunate enough to be selected to present a session at the Society for Cable Telecommunications Engineers (SCTE) show in Denver this week. The event is the biggest cable-focused one of the year, and to get a chance to present to an audience of distinguished operators from around the world was completely due to the critical nature of the topic of my presentation: Maximizing QoE to the Home in a hybrid DOCSIS environments. This is a topic of great interest to cable operators, not just because of some of the unique challenges presented by DOCSIS environments, but simply because all operators want to know how they can deliver a high QoE to their subscribers while minimizing costs. Broadband subscribers are becoming more educated and more aware of the quality of their broadband, and they want it to get cheaper even though they want more bandwidth over time.
The Importance of the Subscriber Experience
Why is the Subscriber Experience important? Have you ever tried to watch something on TV and the service went out? Tried streaming Netflix and you kept buffering or got that “Title not available at this time” message? Had a choppy voice/VOIP call? These experiences are what cause customers to seek alternate providers – and the cost to re-acquire a subscriber is greater than the cost to gain that subscriber initially. The opening few slides of my presentation basically presented this as the dilemma for the MSOs – All networks will eventually get congested – and how the operator manages that congestion will determine the subscriber experience for their customers. The recent dust-up between Netflix and several operators in the US that gave the error message “your operator’s network is slow" (I am paraphrasing here of course) was a highly visible incident, and it pointed directly to the source of a great deal of pain for the operators. The subscriber will almost always blame the service provider when they have problems with broadband traffic.
Managing Network Congestion and delivering Service Plan Assurance: If you have a hammer, everything looks like a nail!
Managing network congestion has multiple levels of sophistication, each resulting in different results. Many operators use the built-in DOCSIS capabilities of their CMTS solutions, and that can deliver some level of fair usage on a per IP address basis. As the saying goes, “If all you have is a hammer, everything looks like a nail!” - and this is the case with this solution. Simple fair usage results in worse performance across all applications, and since video traffic needs different management techniques than web traffic, brute force packet queuing techniques that do not have application and service plan awareness often degrade the quality of service for subscribers for today’s applications almost as bad as no management.
Service Plan Assurance: Fair Usage, Service Awareness, and Transparency
Our solution for managing congestion is called Service Plan Assurance, and combines Fair Split (a fair usage technology leveraging Blue and Stochastic Fair Blue - SFB), Fair Factor (Weighted Active Queue Management with Service Plan Awareness that supports bandwidth borrowing), and full transparency. In a Service Plan Assurance deployment, the network can deliver a higher quality of experience to the subscribers with the Procera solution. First, the network operator can ensure that bandwidth is divided evenly among active subscribers during times of congestion (no enforcement when the network is not congested). In addition to the fair usage technique, SFB queuing tracks the reaction to queuing for each flow, and if the flow does not react to congestion management techniques, it will be managed differently than well behaved flows. Secondly, Fair Factor allows the operator to distinguish between different bandwidth plans that have been sold. A 5Mbps subscriber should not necessarily have the same bandwidth available as a 100Mbps subscriber, so Fair Factor allows for programable ratios for the different active service plans on the network. If a high bandwidth subscriber is not using their portion of allotted bandwidth, it will be made available for the rest of the active subscribers in a fair ratio (i.e. Borrowing). Finally, any actions taken as a result of congestion management are captured and available in real-time or through historical reports for the operator (who can also provide directly to the customer or to regulatory bodies monitoring the subscriber quality of experience. This is a big differentiator for the operator – they can immediately determine what subscribers were affected by congestion management, what applications were affected, and the overall performance of their network with the delivered QoE. There are movements around the world to monitor this on broadband networks, so operators need this information to be able to respond to customer and regulatory inquires. Below are two examples of reports that resulted from our network testing.
Documenting Fair Usage
This report below shows the packet drops using different queuing techniques. As you can see, there appear to be fewer drops in the standard queuing technique, a spike in drops using SFB, and then one group has almost no drops (due to Fair Factor and Fair Split).
This is bad right? Well, it turns out that it is not bad after all. In the first period, the packet drops are not sophisticated, so the applications that are being affected react poorly to the management, and the network performance suffers. The chart below show the same time period, but measures the latency of the traffic.
As you can see, the latency in the final two time periods is dramatically lower than the first period due to the active queue management used. Voice and Video applications will befit greatly from this, as will social networking applications that have a great deal of interactivity.
Our goal is to enhance the subscriber experience – and our Service Plan Assurance solution is just one of our methods that we are deploying with network operators worldwide. We will have a whitepaper on this topic posted soon, so keep an eye out for it!