Improve Workforce Productivity and Collaboration with Integrated Voice and Video Optimization for SD-WAN
November 14, 2017
It’s been an exciting 2017 for Versa Networks with many new customer wins (Comcast Business, Verizon, KDDI, Singtel, S-Net Communications, and others) and the release of an extensive list of new features (e.g. Wi-Fi, embedded switching, LTE and 3rd party VNF support capabilities) enhancing our portfolio to software-define more than just the WAN (SD-Branch). Today we announce support for enhanced unified communications optimizations by supporting the native analysis of over 150 voice and video codec types to create live, in-situ, MOS (Median Opinion Score) calculation reporting.
The Versa Cloud IP Platform has always been able to identify voice and video traffic to dynamically and intelligently steer the traffic across the WAN but the support for codecs and MOS based scoring, with analytics and associated traffic engineering capabilities takes this to a whole new level.
Standard SD-WAN solutions can simply identify a voice or video packet, abide by a policy which is defined by loss, latency, jitter and delay network metrics along with circuit priority (at best) to steer the traffic along in an optimal fashion. However, what most cannot do is give a real-time assessment of the actual (end-user) experience of the UC session AND provide visibililty of the actual experience of the session as captured by the MOS score on a per-session and per-direction basis.
This is important because in standard solution offerings the metric monitoring is of the circuit (transport) and how well the network path is performing. SD-WAN vendors may also define QoS policies to prioritize voice or video over data, pretty standard SD-WAN stuff. However, they aren’t making decisions based on the quality of the actual application/end-user experience, hence it’s more of a decision based on the quality and behavior of a set of circuits or transports – specifically the overlay with some passive monitoring of DIA (direct internet access) which is mainly again just watching for general retransmissions or packet loss.
But how do you support and ensure that the actual end-user experience is being maintained when traffic exits the overlay? How do you even know if the user is having a good experience on their phone call or video conference?
Not many users call IT to say “Hey, that phone call I was just on, sounded GREAT! Thank you” – it’s usually a “Hey, my calls keep dropping!” or “We had to reschedule our video conference because the video feed kept cutting out”. This creates a reactive feedback loop that is geared towards a negative experience which is missing significant data.
Well luckily there is a way now! The Versa Cloud IP Platform does just that. It monitors both voice and video traffic (regardless of codec ), breaks the session up into call legs and calculates an ITU compliant MOS rating and makes traffic decisions based on the actual MOS score of the call(s).
This is important to network operators because the relationship between network performance metrics (delay, loss, jitter, etc) and application experience is pretty complicated.
Here’s an example: it’s highly possible that even if packet loss and jitter metrics individually are well within a defined threshold that there could be serious degradation of the experience of an application. The combination of both low packet loss and higher jitter often results in service degradation perceived by the user.
How can you correlate network metric and user experience? That’s where MOS comes in. Tracking to a single universally accepted metric provides a method to measure the interaction between the various underlying network performance metrics and present it in a simple and well understood format to represent the application end-user experience. This provides a correlation of the calculated user perception and actual experience of the application – voice or video in this case.
This is better for a number of reasons actually. Imagine an enterprise with 100 sites, all with MPLS and broadband. Each site is securely connected via tunnels over both transports, along with local breakout policies for non-internally destined traffic. Standard practice says to: prioritize voice and business class video over all else, use MPLS as primary path and broadband for secondary. Now let’s say you’re making a video call on your internal network and policy says to use MPLS as primary path. Makes sense because of the inherent SLA and QoS on that transport. Now let’s say the MPLS transport isn’t experiencing any loss, abnormal delay or jitter – but the video call isn’t going very well – its choppy and pixelated. What triggers an action to correct this or even look for a better path? How would you even know something isn’t right unless a call or email comes through with “Urgent: Exec video call failing. Bad times. Need help now”? In summary, how can you ensure even within your network that the network is providing a positive user experience for critical services like voice and video?
With built-in codec support and the ability to determine and update MOS data in real-time, the network now can do something about it. When combining this with SD-WAN SLA policies like: if MOS <4.1, do something – (e.g. change transport, move other traffic off, replicate voice, and more) then the network can now respond to the experience and also provide real-time feedback on the effectiveness of those changes. The point here is using MOS as the metric to understand and convey the end-user experience and determine how and when to modify operations is really the key benefit. Again, the MPLS transport is great from the caller but what if the path to the receiver wasn’t? What if the receiving side had a better path with direct internet? Meaning, the overall metrics associated to the MPLS transport were not identifying SLA violations, but the quality of the video traffic was poor as identified by MOS – so how would you signal a change? Why not use a better path? Well to be a broken record, you can normally of course – but you’re waiting on being told that the actual transport is suffering, not the application (voice/video), so you wouldn’t, in this scenario.
With MOS reporting and correlation, you can. You can visualize the end-user UC experience and quality in real-time for voice and video and act on it – making sure that dreaded phone call to IT doesn’t even have a chance to happen.
Reporting is another reason why this is great. You’re asked – how well is the WAN performing? How well are the actual applications doing?
That last question would be tougher to answer with vanilla SD-WAN reporting because you don’t know the actual session quality user experience. In the absence of such knowledge, you rely on the reporting of your IT tickets and reports from your on-premises UC solution or hosted/managed solution. MOS reporting is valuable – it enables you to see and correlate voice and video experience directly within the Versa SD-WAN solution through the single-pane-of-glass.
Another great reason is that with codec support and MOS based policies, we can make SBCs more efficient. Whether they are local or cloud hosted, we can ensure properly mapped policies and service chain more effectively for any calls that are terminating on the SBC. Couple that with Versa SD-Security and things like CGNAT, QoS and security (ie: stateful services) can be offloaded so the SBC (premises or cloud based) can have more resources to do what it’s best at – dealing with calls.
In quick summary, the inherent support for multiple voice and video codecs and MOS based policies and reporting enhances the Versa Cloud IP platform to deliver a more optimized Unified Communications experience. This is whether it is on-premises based UC or cloud/managed (UCaaS) like our partner S-NET Communications, and VergX resellers. The end goal is to more accurately identify applications, correlate them to their device, location and user, while really understanding the end-user experience of the application to take action to keep that experience positive.