Skip to content

Chat Message Latency KQL¶

Analyze chat message delivery times and identify common performance bottlenecks.

Query Description¶

This query retrieves recent chat message sent events, filters for those with latency, and summarizes the average and maximum latency for each thread and sender.

KQL Query¶

ACSChatMessageSentEvents
| where TimeGenerated > ago(1h)
| where ResultType == "Succeeded"
| extend Latency = datetime_diff('millisecond', TimeGenerated, StartTime)
| summarize 
    AverageLatency = avg(Latency), 
    MaxLatency = max(Latency) 
    by ThreadId, From
| order by MaxLatency desc

Explanation¶

Field	Description
`TimeGenerated > ago(1h)`	Filters results to the last hour to focus on current issues and improve performance.
`ResultType == "Succeeded"`	Selects only messages that were successfully sent.
`datetime_diff('millisecond', TimeGenerated, StartTime)`	Calculates the latency in milliseconds between the message start and completion times.
`summarize AverageLatency = avg(Latency)`	Groups the results and calculates the average and maximum latency for each thread and sender.
`by ThreadId, From`	Groups the results by thread ID and sender identity.

Insights¶

Observed Latency: Look for average latency greater than 500ms, which may be noticeable to users.
Performance Bottlenecks: High latency for a specific thread or sender may suggest a network or device issue.
Volume Analysis: A high count of messages with latency may suggest a service-level issue or heavy load.

See Also¶

Sources¶

Azure Chat Diagnostic Log Reference