One key difference between IP data and voice or video over IP (VoIP) is that with data the amount you can dump on your network seems limitless. Sure, the more data you dump, the slower the overall response will be for your users, but the degradation is linear and oftentimes goes entirely unnoticed, especially if the data dump is short-lived. This is certainly not the case for VoIP. In the case of VoIP, the addition of one call can bring down the quality for all the calls on the system; the same can be said with the addition of streaming just one more video.
Why? Voice and video data are very time sensitive and have latency thresholds. When you reach the latency thresholds, the real-time nature of the data renders the delayed data essentially useless. If the data does end up arriving, once it is too late to insert it back into the correct order in the data stream the data will just be dropped and the quality of the video or voice stream will be degraded. Even if all your calls and videos have been working fine, adding just one more call can ruin them all by tipping the latency budget for your VoIP QoS queue, which services all of your VoIP traffic, over the edge.
As companies incorporate more voice and video over IP technologies on their network and employees stream more videos for entertainment, it is important to have a full understanding of the pervasiveness of video and voice on the network. A recent report by Cisco revealed that in 2010 global Internet video traffic surpassed global peer-to-peer traffic, and Internet video to TVs tripled in 2010 and will increase 17 – fold by 2015. (To read the full white paper, check out this link.)
We’ve covered in detail some of the challenges of delivering video over IP in our post IP Video – It’s like Living with a Teenager, but one other element of video that causes a challenge is the need to have it work in tandem with voice. It’s not just a video signal, but a highly correlated voice signal that goes along with it. Thus, you need to take all these unique characteristics of video and couple it with the challenges that are faced with simple voice over IP, which include jitter (when packets are not delivered with a specified spacing), latency, and packet loss – the three headed monster of real-time protocols. To learn more about combating these tedious issues, check out our blog on the subject.
As more companies move towards a shared ecosystem and begin using SaaS and cloud computing, real-time protocols like VoIP and video could be adversely affected as overall network latency increases, and the real-time protocols, given their queuing priority, may also affect application performance more so than in traditional network architecture with centralized data centers. Before making any significant changes to a network, it is critical that you understand how all of the current network traffic is performing and interacting, including IP data and real-time applications. This is done by performing baseline assessments of your network. With solid baselines in place, you can easily determine the impacts of adding new applications or increasing real-time protocols, as well as whether or not a move to the cloud is advantageous or not.
Video and VoIP are not going away any time soon. In fact, real-time protocols, especially video, is only going to increase at ever increasing rates. More high-priority traffic like VoIP and video means even greater delays for your existing enterprise application data, which may already be under stress as you move to cloud or service-based applications. Understanding and possibly limiting video traffic to only that which is mission critical for your business may become a reality. Or maybe just a hop to 40, and or even 100G will do the trick, but that’s an expense alternative to address a proliferation of traffic which shouldn’t be on your network in the first place.