With Client Analytics, we are referring to any metrics or KPIs that help provide a better understanding of the end-user Quality of Experience (QoE). This becomes critical when there are playback issues to troubleshoot or for establishing alerts when monitoring rules meet a particular threshold.

• Troubleshooting Client Playback
• Monitoring Client Analytics

Troubleshooting Client Playback Issues

When troubleshooting playback issues it can be helpful to first verify that the problem is not originating from the broadcast source.

Questions to consider:

• Do all client apps experience the same issue?
• Is the problem observed regardless of browser or platform being used?

If the answer is yes to either of those questions, that may indicate a broadcast problem. You can use the Millicast Interactive Multiviewer demo as a test reference for confirming the client is able to view playback.

👍

Broadcast Monitoring

For broadcast issues, the Live Monitoring tool is available from your account Streaming Dashboard or by using the Broadcast Monitoring REST API to get more insight into what may cause problems with a stream.

Diagnostics Report

When a broadcast is healthy, there can still be scenarios where an individual playback client may experience poor player performance due to regional network disruptions, app misconfigurations, client platform versions, codec compatibility, etc.

To help facilitate investigating out why a client is experiencing a degraded experience, there is a utility function available from the Client SDKs called diagnose(). This method gathers up pertinent details such as:

✓ The version of the SDK being used.

✓ User agent of the browser and/or platform.

✓ Connection status, playback cluster, and stream account details.

✓ WebRTC stats such as round-trip time, bitrate, packet rate, mime type, resolution, jitter, etc..

It is common to need this type of context from your application when an end user reports a problem. This level of insight can be valuable when shared with our support team to quickly find a resolution when troubleshooting.

How-to Diagnose with the Web SDK

To generate a diagnostic report, call the diagnose() method.

let report = Logger.diagnose(config);

The result is a JSON object that includes a snapshot of what was happening on the client as observed in SDK logs and statistics. When building your application, a button or menu item to report an issue that sends this data is a common pattern.

There are some options that can be used with the config to further customize the resulting report.

• By default the stats for the prior minute collected every 1s will be included in the report. By modifying statsCount this can be customized.
• A large number of logging events will be buffered and included, but this number can be extended or shortened by using the minLogLevel and historySize to indicate which logs to include and how many of them. The default is TRACE but can be changed to INFO or DEBUG to increase or decrease the amount of detail respectively.

let config = {
  statsCount: 120,
  historySize: 200,
  minLogLevel: 'DEBUG',
};

See the Data Dictionary below for examples of some of the data returned.

Ongoing Monitoring of Client Analytics

It is common practice to proactively monitor all clients for indicators of streaming issues. The Client SDKs provide access to a stats event that emits a snapshot of WebRTC indicators at a frequent interval. This data can be used in combination with an Application Performance Monitoring (APM) solution or Observability tool to alert and integrate with other third-party analytics services for a complete picture of the end-viewer experience.

Analytics Collectors

There are third-party Client Analytics integrations available that can serve as a collector or monitoring tool for storing analytics.

• Datazoom
• Datadog

The same pattern can be used with other tools such as Grafana, Splunk, New Relic, etc.

👍

Viewer Reporting Analytics API

The Viewer Reporting REST API is a premium Analytics APIs option made available to enterprise customers. It offers key insights from our server logs to help identify viewers, location, and other metrics. If you would like to use this endpoint, please submit a support ticket or reach out to your sales/solutions contact.

How-to Configure Stats Collection Frequency

The default behavior is to emit a stats event every one second (1s). To manage bandwidth and data storage costs, this value can be tuned to meet any observability needs at higher or less frequent intervals.

// Reduce stats collection frequency to be every 5s instead of the default 1s
const option = {
  peerConfig: {
    statsIntervalMs: 5000
  }
}

await millicastViewer.connect(options);

Data Dictionary

The data measures and metrics gathered for diagnostics are described below.

Software Identification

Client Environment

Client Connection

Content Characteristics

Playback Performance