We consistently get calls from folks looking to solve voice or video over IP (VoIP) problems. Even if your company is still in the age of the dinosaurs using PBX equipment from 1987, there’s still VoIP traffic on your network, most probably in the form of integrated VoIP sessions with WebEx meetings and Skype. So, regardless of your telephony technology, you have to deal with VoIP problems, and in our experience it’s usually not the VoIP application or technology that’s to blame – it’s the underlying network.
Network misconfiguration is a typical culprit in VoIP problems, as are unforeseen or increasing network/processing latencies, excessive VoIP traffic, and conflicts between traditional network data and real-time network traffic.
Luckily, these problems have an easy remedy: equip yourself with a network analysis solution that treats VoIP like any other data type, allowing you to assess all traffic on your network, simultaneously. The solution must include summary-level dashboards to provide a baseline of your network activity prior to implementing VoIP, as well as granular analysis to help you assess more specific problems once VoIP is implemented.
So let’s break down these two steps for a somewhat more detailed view of what we expect from our network analysis solution when it comes to VoIP.
If you haven’t already deployed a VoIP solution, this step is a must. And if you already have a VoIP solution in place, baselines still play a critical role in your overall network analysis. Developing baselines is as much a process as it is an ongoing activity. Before making any baseline measurements, decide what is important to baseline on your network. Typically, overall utilization over time and identification of top talkers (network bandwidth users) and applications are a table stake, but depending on your network architecture other data points may also be essential, like the overall utilization of specific servers in the data center or the traffic hitting your WAN link. Once you’ve identified the critical data points for your network periodically run tests which run over long spans of time to monitor these points, recording the summary level statistics. Then establish a routine for performing the same checks on a periodic basis so you’ll have an apples-to-apples comparison of your network performance over time. If you haven’t deployed a VoIP solution yet, this information will be invaluable in helping you to determine if your current network infrastructure can handle the predicted increase in network traffic due to the VoIP system. And don’t forget periodic measurements of metrics like latency and packet loss. Poor performance in these areas can obliterate VoIP performance, so you definitely want a baseline of these metrics as well before deploying VoIP.
Analyzing Specific VoIP Problems
Once VoIP is deployed, one of your key strategies should be to continue your baseline measurements. This will help you identify many issues before they become true problems, including common misconfiguration issues like QoS not being appropriately configured across the network path. But equally important is that it will help identify any areas that are exceeding design specifications. Perhaps your design assumed that no more than 80% of your VoIP users would be making calls simultaneously, but through baseline assessments you identify periods of time where this percentage is exceeded. If your overall design margin is thin, oversubscribing your predicted VoIP usage can cause issues for all VoIP users, not just the few who cause the limit to be exceeded.
Analyzing issues at the call level also becomes critical after deployment, as this is most likely the way problems are going to be reported to you. If a specific user is reporting overall call quality issues, it could be an isolated issue, or it could be an indication of overall trouble on certain network segments. Using a network analysis solution that treats VoIP like any other network data allows you to see everything that is happening on the network while the problem call was active. Perhaps a large data transfer was initiated while the call was active, causing network latency to increase and resulting in an unacceptable increase in dropped packets. Or perhaps the call routing has changed and the new gateway has QoS misconfigured. With a full-featured network analysis system with VoIP analysis capability these issues, and many more, will be easy to identify.
It’s a Commitment
Activity on your network will change over time, so constant vigilance in monitoring your network is a must. And these activity changes will likely affect VoIP performance, so having one analysis solution that monitors everything together will ease your workload and your troubleshooting efforts. And keep in mind that even though it’s awfully tempting to blame the VoIP technology, odds are any performance issues are network-based, so roll up your sleeves and start your investigation there.