Quality monitoring

更新时间:
复制 MD 格式

Monitor the quality of your ApsaraVideo Real-time Communication service with application-level metrics such as Video Stuttering Rate, Audio Stuttering Rate, and 400ms Latency Compliance Rate. Channel analysis, call analysis, and remote subscription quality analysis help you identify and resolve issues quickly.

Overview

The quality monitoring dashboard provides application-level data monitoring and analysis. Use it to assess audio and video service quality, validate new version stability, or identify widespread stuttering issues.

Use cases

  • Development and debugging: After you integrate an SDK, verify that the quality of your audio and video streams meets expectations.

  • Version validation: Before and after a new version release, compare key quality metrics to assess the version's stability.

  • Issue diagnosis: When you receive widespread reports of stuttering or connection failures, use this feature to investigate trends and isolate the problem.

  • Trend monitoring: Track the overall health of your application's quality over time during regular operations.

Prerequisites

Before you use quality monitoring, make sure the following requirements are met:

Procedure

  1. Log on to the ApsaraVideo Live console.

  2. In the left-side navigation pane, choose Live + > ApsaraVideo Real-time Communication > Quality Monitoring.

    image

    Note

    Select an ApsaraVideo Real-time Communication application ID to view the corresponding quality data. Each data point represents the average value over a 5-minute interval. You can find your application IDs and names on the Live + > ApsaraVideo Real-time Communication > Application Management page in the ApsaraVideo Live console.

Global overview

The Overview tab displays key quality metrics for your ApsaraVideo Real-time Communication application. Filter by the following dimensions:

  • Client Type: Native or Web. The default is Native.

  • System: All Systems, Android, or IOS. This filter applies only to the Native client type.

  • Time Range: Query data for a custom period from the last hour to the last 3 days. The default view shows data from the last hour.

Notes:

  • The 400ms Latency Compliance Rate metric is available only for data generated after February 1, 2025.

  • Data is plotted in 5-minute intervals, where each point represents the average value for that interval. You can hover over a data point to view its timestamp and value.

The metrics on the Overview tab are defined as follows:

Metric

Description

Success Rate of Connection within 5 Seconds

The percentage of attempts to join a channel that succeed within 5 seconds.

First Frame Duration

The time from a subscribe request to the rendering of the first frame, measured in milliseconds (ms).

400ms Latency Compliance Rate

(Total duration of compliant video transmission / Total video usage duration) × 100%. A transmission is compliant if the end-to-end latency from the sender to the receiver (excluding jitter buffer delay) is less than 400 ms.

Audio Stuttering Rate

(Total audio stuttering duration / Total audio usage duration) × 100%.

Video Stuttering Rate

(Total video stuttering duration / Total video usage duration) × 100%.

Online Users

Peak online users: The maximum number of concurrent users within the specified time range, which is the peak count of unique UserIDs online at the same time.
Total active users: The total number of unique users during the specified time range, which is the count of distinct UserIDs.

Online Channels

Peak online channels: The maximum number of concurrent channels within the specified time range, which is the peak count of unique ChannelIDs active at the same time.
Total active channels: The total number of unique channels during the specified time range, which is the count of distinct ChannelIDs.

Channel analysis

The Channel Analysis tab on the Quality Monitoring page provides real-time rankings, historical metric rankings, and call records by channel. It contains three sub-tabs: Real-time Call Ranking, Metric Ranking, and Call List.

Filter by the following dimensions:

  • Application: The first application is selected by default.

  • Client Type: Native or Web. The default is Native.

Real-time Call Ranking

This tab ranks online channels by their quality metrics from the last 30 seconds, helping you quickly identify abnormal channels.

  1. On the Channel Analysis page, select the Real-time Call Ranking tab.

  2. At the top of the page, use the quick navigation links to jump to the ranking table for a specific metric: Online Users, 5s Join Success Rate, Time to First Frame, 400ms Latency Compliance Rate, Audio Stutter Rate, or Video Stutter Rate.

  3. Each ranking table displays up to 20 channels, showing the channel ID, start time, and relevant metric values. Click Call Analysis in a row to open the Call Analysis page for that call.

Note
  • Online users: The number of users currently online in the call.

  • Active users: The total number of unique users who participated in the call, which is the count of distinct UserIDs.

  • Ranking logic: Channels are ranked from high to low for Online Users. For 5s Join Success Rate and 400ms Latency Compliance Rate, channels are ranked from low to high (worst-performing channels first). For Time to First Frame, Audio Stutter Rate, and Video Stutter Rate, channels are ranked from high to low (worst-performing channels first).

Channel Metric Ranking

The Metric Ranking tab displays historical quality rankings by averaging metric values at the channel level over a selected time range.

  1. On the Channel Analysis page, select the Metric Ranking tab.

  2. Select a Time Range. The default is the last hour. You can select a custom time range of up to 24 hours within the last 30 days. The minimum statistical granularity is 5 minutes.

  3. The page displays channel rankings for metrics such as 5s Join Success Rate, First Frame Duration, 400ms Latency Compliance Rate, Audio Stuttering Rate, and Video Stuttering Rate. The ranking logic is the same as in the real-time call ranking. This is a view-only page and does not support navigation.

Call List

The Call List tab lists call records within the selected time range that match your filter criteria. Use it to find a specific call and open its analysis page.

  1. On the Channel Analysis page, select the Call List tab.

  2. Select an Application and a time range (defaults to today; custom range up to 30 days). You can enter a channel ID in the channelId field to search for a specific channel.

  3. Click Call Analysis in a row to open the Call Analysis page for that call.

Note
  • Active Users: The number of unique users who participated in the call. A user who joins and leaves multiple times is counted only once.

  • Peak Online Users: The maximum number of concurrent Online Users during the call's duration.

Call analysis

Click Call Analysis from the Real-time Call Ranking or Call List tab to open the call details page. This page includes three sections: Quality Overview, Metric Ranking, and User List.

Quality Overview

The Quality Overview section shows quality trends for the call within the selected time period.

  1. Select a Client Type (Native or Web). The default is Native.

  2. Use the timeline to select the time period you want to view. The timeline displays the call's start and end times, and the query interval is fixed at 1 hour: 

    • If the call has ended, the timeline defaults to the last hour before it ended. 

    • If the call is ongoing, the timeline defaults to the first hour after it started. 

    • If the total duration is less than 1 hour, the timeline cannot be adjusted.

  3. The chart displays the following metrics at a 30-second granularity: Online Users, 5s Join Success Rate, First Frame Duration, Audio Stuttering Rate, Video Stuttering Rate, and 400ms Latency Compliance Rate. The definitions of these metrics are consistent with those in the Overview section. For more information, see the metric table in the Global Overview section above.

Metric Ranking

The Metric Ranking section ranks users within the call based on their quality metrics, helping you identify users with poor performance.

  1. This section provides ranking lists for metrics such as 400ms Latency Compliance Rate, Downlink Packet Loss Rate, Uplink Packet Loss Rate, Audio Stuttering Rate, and Video Stuttering Rate. Use the quick navigation links at the top to jump to a specific metric's section.

  2. Each list displays the join time, leave time, and metric value for each user. The 400ms Latency Compliance Rate list shows the 50 worst-performing users (lowest success rate). The packet loss and stuttering rate lists show the 20 or 50 worst-performing users (highest values).

  3. Click View Details in a row to open the User Details page for that user.

Note

Upstream/downstream packet loss rate: (Number of packets lost in a time unit / Total number of packets in a time unit) × 100%.

User List

The User List section displays all users in the call and their status. Filter by User ID or publisher ID, and navigate to individual user details or remote subscription quality pages.

  1. In the User List section, you can enter an ID in the UserId field to search for a specific user. You can also enter an ID in the Publish UserId field to find users who have subscribed to that publisher.

  2. The Subscriber count indicates the total number of unique users (by UID) who subscribed to this user's stream. Clicking this number automatically populates the Publish UserId field to filter the subscriber list.

  3. Actions: Click View Details to open the user's call details page. For users who have published a stream, a Remote Subscription Quality button is also available. Clicking it takes you to the Remote Subscription Quality tab on the user details page.

Note

List sorting priority: Online users are listed before offline users. Among users with the same status, publishers are listed before subscribers. Users are then sorted by join time in descending order.

User details (formerly Single-point Troubleshooting)

Click View Details or Remote Subscription Quality from the Metric Ranking or User List sections in call analysis to open the user details page. This page shows the user's publishing quality, subscription quality, and (if applicable) remote subscription quality.

The user details page contains the following tabs:

Tab

Description

Stream Ingest Quality

Formerly Push Quality. Displays push stream data, data received by Alibaba Cloud nodes, and upstream packet loss rate. The upstream packet loss rate chart was moved to this tab from the Endpoint Load and Network tab.

Stream Pulling Quality

Formerly Pull Quality. Displays data sent from Alibaba Cloud to the user, and a downstream packet loss rate chart.

Remote Subscription Quality

This is displayed only for users who have performed a publish operation. It shows the overall quality for remote users who subscribe to this user's published stream. The data is collected at a 30-second granularity on a draggable timeline that spans 1 hour. Metrics include the number of subscribers, First Frame Duration, Audio Stuttering Rate, Video Stuttering Rate, and the 400 ms latency compliance rate. The definitions are consistent with the Global Overview.

Device Load and Network

Displays endpoint CPU and memory usage. The upstream packet loss rate chart has been moved to the Stream Ingest Quality tab. For Web clients, the CPU and memory charts do not appear if data is unavailable.

Events

Records all system and user events within the channel, including joining a channel, starting to publish, re-publishing, and subscribing to a stream.