Fotolia
How can you measure WebRTC performance?
Organizations can take two approaches to measure WebRTC quality, both of which assess real-time communications performance on end-user devices and the network.
Part two of this ask the expert response discusses the different ways to measure Web Real-Time Communications performance. In part one, our expert Tsahi Levent-Levi explains what factors influence WebRTC performance.
WebRTC performance quality can be measured on the device side or the server side. Traditionally, voice over IP applications are measured on the server side. Real-Time Transport Control Protocol and Real-Time Protocol packets are collected from the media servers. Quality scores are also measured based on bit rate, jitter, latency and packet loss. These offer a fairly good picture of network performance as a whole and of the individual sessions.
Often, the collection of these metrics takes place directly on the device through WebRTC's statistics. In some cases, media doesn't traverse through a media server, but rather runs peer to peer between two devices. When this occurs, there is no media server to latch onto to collect metrics.
Two methods to measure WebRTC performance
WebRTC performance can be measured in two ways: on live user sessions or on predictable synthetic traffic.
Measuring live user sessions enables you to understand the perceived quality your users are currently experiencing. If you are measuring metrics collected from devices, you can use third-party options, such as callstats.io, or develop tools in-house using the RTC stats open source library. Collecting metrics from media servers can be done using commercial tools such as Homer.
The second approach to evaluating WebRTC performance measures predictable synthetic traffic. For this approach, you use devices in different locations on the network and simulate user traffic to measure session quality and where particular issues may arise on the network. The idea here is that you can observe changes in the traffic behavior that are directly linked to your infrastructure, giving you more actionable information beyond the end-user experience. If you plan on measuring WebRTC performance using predictable synthetic traffic, you can use a third-party offering, such as testRTC.
Not measuring quality can prevent you from understanding and resolving quality issues for your users and customers. Measuring both live user sessions and predictable synthetic traffic will give you the best visibility for your users and network infrastructure.
Do you have a question for Tsahi Levent-Levi or another expert? Ask your enterprise-specific questions today! (All questions are treated anonymously.)
