Telemetry for diagnosing poor call quality typically starts with the unified communications as a service (UCaaS) provider (like Microsoft Teams or Zoom) who supply dashboards with call quality metrics. These call quality metrics include the mean opinion score (MOS) which is a numerical measure of the human-judged overall quality of voice and video sessions, reported on a scale of 1-5, with 5 being a perfect call. But that’s where the diagnostics often ends––finding where a user with a 3.6 MOS call quality (akin to a garbled cell phone call) actually had the breakdown is very difficult.
Finding the root cause of the breakdown requires detailed network path analysis to identify sources of packet loss or latency, and the user endpoint diagnostics to identify CPU, memory, or Wi-Fi stats gone awry. Even if these data points are available, they are typically captured post-incident, so there is no way to correlate what the UCaaS vendor is reporting with statistics captured during the Teams or Zoom meeting in question.
There’s finally some good news to those struggling with these issues.
Figure 1: ZDX ingests API call quality data from Microsoft Teams and Zoom and marries this with application, network, and device performance metrics
Zscaler’s own digital experience monitoring offering, called Zscaler Digital Experience (ZDX), has been helping clients diagnose call quality problems since its inception.
But now ZDX takes it to the next level, with new integrations with the Microsoft Team Call Quality API and the Zoom API to pull in call, video, and sharing quality stats for every Teams or Zoom meeting taking place. This data is then seamlessly married with network path (CloudPath) and endpoint device metrics that ZDX has been collecting during the course of the meeting (from every employee every few minutes) (see Figure 1).
When a user reports a poor Teams or Zoom experience, ZDX provides an integrated workflow to look at call quality metrics, network details, and endpoint details, helping to identify the root cause of the problem, which could be anything from weak Wi-Fi, high ISP latency, high CPU, and more (read my blog on tips for finding root causes here).
For Zscaler customers, ZDX integrates with your existing deployment of the Zscaler Client Connector to provide this capability. No new agents are required. And, because the ZDX-enabled Client Connector is always on, network and endpoint metrics are continuously being collected. This offers a wealth of forensic evidence during the actual incident.
Now, when someone complains that they ‘can’t hear you’, the reason why is only a few clicks away.
Here’s how it looks in practice in ZDX:
Figure 2: ZDX computes an aggregated score based on call quality metrics gathered from the UCaaS provider
Figure 3: ZDX displays this call quality score over time and on a world map, to identify global, regional, or local issues that may be occurring
Figure 4: ZDX captures the meetings that occurred over the specified time, along with the associated call quality, ZDX Score, and participant information
Figure 5: For each meeting, ZDX can display all of the participants of that meeting, along with their individual call quality, ZDX score, audio, hardware, and connection information
Figure 6: ZDX highlights audio, video, and sharing latency, jitter, and loss metrics for each participant in the meeting
Figure 7: While the Teams or Zoom meeting is in progress, ZDX captures CloudPath details and can pinpoint spikes in latency or packet loss and the specific hop where that spike occurred
Figure 8: While the Teams or Zoom meeting is in progress, ZDX also captures changes in the user’s device, highlight differences in network settings, CPU speed, etc., that may be impacting call quality