By David,
Microsoft publishes Azure's region-to-region latency in its "Azure network round-trip latency statistics": the median (P50) round-trip time between Azure regions, measured over 30 days on Microsoft's backbone and updated every 6 to 9 months. It doesn't publish latency from your users to Azure, which is the number that matters for Azure Virtual Desktop. For that, Microsoft's guidance is a round-trip time under 150 ms from the client's network to the host pool's region, and the way to measure it is AVD's own connection data.
The short version: Use Microsoft's latency statistics to compare regions with each other, for example a primary and a disaster recovery region. Use the WVDConnectionNetworkData table in Log Analytics, or AVD Insights, to see the round-trip time your users actually get. Aim for under 150 ms; Microsoft says above 200 ms "might affect user experience." Put session hosts in the region closest to your users, and use RDP Shortpath for more consistent latency.
This expands the latency section of our Azure regions directory. Everything below is from Microsoft Learn as of November 2026, with each source linked.
Microsoft's Azure latency statistics
From Azure network round-trip latency statistics:
- What it measures. Round-trip time between Azure regions, from "internal network probes that continuously monitor the performance of the Azure backbone network", collected every minute.
- What it publishes. The median (P50) round-trip time in milliseconds over a 30-day window. The current data covers the 30 days ending July 30, 2026.
- Direction matters. Values are directional: Microsoft's own example is 8 ms from East US to East US 2 and 9 ms back.
- Coverage and freshness. "Expect an update to these tables every 6 to 9 months. The following tables don't list all public Azure regions."
- What it isn't. It's region to region on Microsoft's network. It says nothing about the path from your office or a user's home to Azure.
When to use it for AVD. To choose a disaster recovery region close to your primary, to place FSLogix Cloud Cache providers or other services near your session hosts, or to check the cost of a session host in one region reaching storage or a database in another.
The latency AVD users can tolerate
Microsoft gives several thresholds, from different pages:
| Source | Guidance |
|---|---|
| AVD prerequisites | Round-trip time "from the client's network to the Azure region that contains the host pools should be less than 150 ms" |
| Connection quality troubleshooting | "Latency up to 150 ms shouldn't affect user experience that doesn't involve rendering or video. Latencies between 150 ms and 200 ms should be fine for text processing. Latency above 200 ms might affect user experience" |
| AVD Insights use cases | Latency should "generally be below 100 milliseconds, with a session broadly becoming of low quality over 200 ms" |
| RDS network guidance | Assistive technology such as Narrator needs "RTT of 20 milliseconds (ms) or better" |
In practice: under 100 ms feels local, under 150 ms is Microsoft's target, 150 to 200 ms is workable for office work, and over 200 ms users will notice. Video and graphics-heavy work want the low end. Microsoft also recommends you "create session hosts in the Azure region closest to your users" (prerequisites).
How to measure your users' latency to AVD
WVDConnectionNetworkData in Log Analytics
With AVD diagnostics sent to a Log Analytics workspace, the WVDConnectionNetworkData table records each connection's estimated round-trip time, EstRoundTripTimeInMs, and available bandwidth, EstAvailableBandwidthKBps. AVD "generates these network data points every two minutes", and the round-trip time "is protocol-agnostic" (connection latency, table reference). Data can take up to 15 minutes to appear (connection quality monitoring).
Median and 95th percentile round-trip time per user over the last week:
WVDConnectionNetworkData
| where TimeGenerated > ago(7d)
| join kind=inner (WVDConnections | where State == "Connected" | project CorrelationId, UserName) on CorrelationId
| summarize MedianRTT = percentile(EstRoundTripTimeInMs, 50), P95RTT = percentile(EstRoundTripTimeInMs, 95) by UserName
| order by P95RTT desc
Look at the 95th percentile, not just the median: a user whose median is fine but whose P95 is over 200 ms is having a bad day several times an hour.
AVD Insights
AVD Insights (aka.ms/avdi) charts round-trip time by gateway region with median and 95th percentile values, so you can see whether a whole office or region is affected rather than one user (Insights use cases).
Before you deploy
The Azure Virtual Desktop Experience Estimator has been decommissioned, according to a Microsoft Q&A answer, with no direct replacement. Microsoft's guidance is to use the region latency statistics and to review them "every two to three months" (connection quality troubleshooting). To test a candidate region with real users, deploy a small pilot host pool there and read WVDConnectionNetworkData for a week.
How RDP Shortpath and the gateway affect latency
RDP Shortpath uses UDP, which "offers better connection reliability and more consistent latency", and with a public network (STUN) path, "the removal of extra relay points reduces round-trip time." When several paths are possible, it chooses "the one with the lowest latency" (RDP Shortpath). The connection information dialog in the client shows whether a session is using UDP (Private Network), UDP or UDP (Relay) (configure RDP Shortpath).
Gateway selection. For connections through the AVD gateway, "gateways are put into groups of 10 ms. The gateway with the lowest latency and then lowest number of existing connections is chosen" (network connectivity).
How Atlas AVD helps
Atlas AVD scores each session's experience from round-trip time and other signals, and in the User Portal, Smart Connect measures latency from the user's own device to each region where their desktops run before recommending one. That shows you the latency your users actually get, region by region.
FAQ
What latency is acceptable for Azure Virtual Desktop?
Microsoft recommends under 150 ms round-trip time from the client's network to the host pool's region. 150 to 200 ms is fine for text work, and above 200 ms might affect the experience.
Where can I find Azure latency between regions?
Microsoft's Azure network round-trip latency statistics page publishes median region-to-region latency over 30 days, updated every 6 to 9 months.
How do I measure my users' latency to Azure Virtual Desktop?
Send AVD diagnostics to Log Analytics and query the WVDConnectionNetworkData table's EstRoundTripTimeInMs, or use AVD Insights, which charts round-trip time by gateway region.
Does RDP Shortpath reduce latency?
It gives more consistent latency over UDP, and with a public network path it removes relay points, which reduces round-trip time.
Is the Azure Virtual Desktop Experience Estimator still available?
According to a Microsoft Q&A answer it has been decommissioned, with no direct replacement. Use Microsoft's region latency statistics to compare regions, and AVD's connection data to measure real users.
Latency data and guidance are from Microsoft Learn as of November 2026: Azure network round-trip latency statistics, AVD prerequisites, connection quality troubleshooting, Insights use cases, RDS network guidance, connection latency, connection quality monitoring, RDP Shortpath, configure RDP Shortpath and network connectivity.