Also in this issue:
Stay Tuned for our Latest White Paper: One-Way Traffic Performance Measurement
Quality of Experience (QoE) is quickly becoming the name of the game in mobile backhaul services. Today’s mobile subscriber expects nothing short of a fixed-line user experience on their smart phone or tablet, regardless of whether they’re downloading a YouTube video, completing a trade transaction, or watching a Netflix movie. And they’re not too interested in hearing service provider woes on how to best deliver these delay-sensitive, bandwidth-intensive applications. In other words, user demand and expectations are driving an increased need for a Quality of Service (QoS) that far exceeds a “best effort” level of service.
The reality of today’s modern mobile networks is that they support multiple applications, each with its own unique performance requirements when it comes to network parameters such as delay, delay variation, frame loss, etc. However, meeting these requirements on the WAN (Wide Area Network) with its bandwidth constraints can be a challenge compared to a Local Area Network (LAN) where bandwidth is much more abundant.
Classifying Ethernet traffic before putting it on the network is therefore imperative in order to properly prioritize different applications across the limited WAN bandwidth and ensure that application-specific requirements are met. And let’s not forget that in the real world, user perception is king. While the user application may not be as sensitive to long delay or delay variation, the user may be sensitive to long wait times.
So, what is traffic classification? In a nut-shell, it’s a technique that identifies the application or protocol, and tags the packets (or just lets them through untouched) based on certain classification policies, which are then used by the network interface device to provide appropriate treatment to those packets.
Research has shown that only 10-20% of traffic is extremely time sensitive, yet right now we’re throwing 100% of the traffic on the same pipe with little or no regard to what’s delay sensitive and what isn’t. For clarification purposes let’s compare this to an airplane where all passengers are treated equally, and boarded in the first-class compartment regardless of ticket price. And, due to the limited space availability in first-class, a mix of first-class, business and economy passengers weren’t able to board that first plane, so you call up another and repeat the same unstructured boarding procedure. And so on. In this analogy where airplane seating capacity represents bandwidth, you’re not only flying with an airplane that ¾ empty, you’re adding more and more bandwidth to accommodate the traffic you’ve left behind, some of which should’ve been on that first flight. This doesn’t make sense.
Traffic classification is critical to optimizing available bandwidth while improving your Ethernet network performance and the user experience. By classifying traffic, you ensure that critical applications such as financial transactions are treated as ‘first-class’ priority and get through as quickly, and as soon as possible. Your ‘business-class’ traffic such as internet browsing, over-the-top or streaming video which may be less sensitive to delay, but more so to delay variation, gets through next. And then you have your ‘economy’ class traffic that still needs to get through, but can probably wait a bit.
Digging a little deeper in how all this works, let’s look at a mobile backhaul network where congestion typically happens primarily in the downstream direction. Ethernet frames originating from the Internet, mobile network controllers, voice gateways, etc. are classified, meaning that a determination is made on the priority class of each frame based on its origin and contents. The frame is switched to the appropriate egress port towards the Ethernet Virtual Connection (EVC) and placed into the appropriate queue for its class. On an ongoing basis, a queue-servicing algorithm takes frames out of the appropriate queue and sends them on the EVC towards their destination.
To continue with the air travel analogy, this mechanism of prioritizing traffic is very much like a lineup to check in at the airport. Rather than having all customers wait in a common queue, higher priority customers (e.g. frequent flyers or business class travelers) are put in a different queue and airline counter personnel service the two queues appropriately so that the higher priority customers do not have to wait as long.
Classifying traffic can therefore make a huge difference in the customer experience for mission-critical time sensitive applications, and can help you optimize the bandwidth you have available. Not to mention leveraging one of the most powerful capabilities of Ethernet, that being the ability to engineer the network in the context of different traffic priorities. For more information about how Accedian Network’s network demarcation devices handle traffic classification, visit the Company web site at www.accedian.com
Understanding why and how one-way traffic performance measurement in wireless backhaul networks can bring essential visibility of performance and help optimize the use of scarce network resources.
Accedian Network’s latest technology white paper reveals that not only do round trip measurements provide incomplete or inaccurate results, they can also provide misleading information that can result in costly over-provisioning, as well as lack of problem identification and real understanding of network performance.
The paper demonstrates how the proper deployment of devices that carry wire rate operational data while correctly measuring application and service traffic performance can bring the correct insight on how a network is functioning. Devices that properly measure uni-directional data flow, aggregate OAM LTE equipment market will reach $17.5 Billion in 2016information while passing data at wire rate will be a critical element in understanding and cost-effectively engineering the networks of the future.
As backhaul networks migrate to Ethernet 4G/LTE and move from single to multiple class of services, these devices will be essential to provide visibility of performance issues that can remain a mystery when only round trip, non-deterministic measurements are used.
The paper examines traffic patterns in mobile backhaul networks and how they impact provisioning and quality of service. It investigates the risks and costs of relying on round trip measurements and then looks at measurement standards and how that knowledge can be used to both engineer and operate the wireless backhaul as it migrates to new standards and usage patterns. Stay on the look out for the complete white paper, coming soon.