In early July, the official version of the “data insight” function of Sonnet Agora Crystal ball was launched. “Data Insight” shows two kinds of data, one is quantity, the other is quality.
The biggest significance of “Data Insight” “usage overview” is that it can help you track the overall trend of audio and video minutes in the past period of time, and provide data reference for reviewing and adjusting business operation strategy. These data graphs are relatively easy to understand.
However, users who see the “quality overview” in the “Data Insight” for the first time may feel just a number of data ICONS in front of them. For a while, they do not know how to recall the quality problems they have encountered and trace the source step by step. So we’re going to take a test case, step by step, from finding the problem, investigating the clues, to finding the basis of the problem, and showing you how to use it.
* Note: The following figure shows the test Demo data
First of all, the page can be divided into three modules: “User experience”, “Join channel” and “indicator analysis”. “User Experience” and “Join Channel” show five quality indicators that affect real-time interactive experience: video lag rate, audio lag rate, network latency rate, login power and 5s login success rate.
“Indicator analysis” shows the data distribution of the above five quality indicators in the dimensions of region, operating system, network type, device type, SDK version and channel scale.
We chose to look at data from June 26 to July 4. The “User Experience” module will display the trend of video lag rate, audio lag rate and network delay rate during this period. What’s more, it automatically filters out data from the worst day and displays it above the graph. As the chart below shows, the worst days for all three indicators happened to coincide with July 1. What happened that day?
We can put the mouse over the data curve on July 1. Click on the data points, and you’ll see two options in a pop-up bubble: “View by hour” and “View distribution.”
To further investigate what went wrong during the day, we clicked View Hour to see the day’s quality data in finer granularity. As shown below, we see that the worst experience is at 21:00.
Next, we click on the 21:00 data point and go to “Channel Data Sampling”. Of course, only data points that meet the “sampling rules” will show “Channel Data Sampling”. For details of the rules, please search ** “Sampling Rules” ** at docs.agora. IO /cn/.
After clicking “Channel Data Sampling”, the sampling details will pop up on the right. The “minutes-video lag rate” data scatter chart at that point in time is listed here. Each dot is a channel. The closer the data point is to the upper right corner, the higher the delay rate of this channel is, and the longer the talk time is, that is, the worse the experience of this channel is.
We can see from the figure below that the data points near the upper right corner are all the same channel.
At this time, if we click any channel number, it will show the number of users (i.e. the caton influence range) and the total video duration (i.e. the caton influence duration) at the time of channel change. At the same time, a “user data sampling” appears.
What is going on with this channel?
Next, click “User data Sampling”, the bottom of the window will show which users are in the channel at that time. As can be seen from the following figure, although different users encounter quality experience problems, the peer users are all the same (red box in the figure). Note that the experience problem may be related to the peer user.
When we click “Call survey” on the right, it will jump to the “Call Survey” function of the crystal ball to query the quality data of the channel at that time in detail.
According to the data, the network status of the sender is normal, but the CPU is abnormal from 18:00 to 21:00. The red data indicates that the CPU usage is too high. When we look at the resolution of video transmission, when the CPU usage is high, the resolution of video transmission is also reduced. But Wi-Fi signal quality is blue, which means the network is good.
Therefore, it can be preliminarily judged that the insufficient performance of the device at the sending end leads to the video lag in this period.
Left left left forecast
The “call survey” has several quality dimensions that can be used to investigate the root cause of a call problem. We will explain the use of the “call survey” by focusing on two typical cases. Stay tuned.