Azure latency check.
Round-trip latency from your browser to Microsoft Azure regions. Hit in parallel, median ms reported. Click rerun for a fresh measurement.
Azure latency check — browser-measured RTT to Microsoft Azure regions using Cognitive Services regional endpoints.Measured live from your browser to 0 Azure region endpoints. Three samples per region in parallel, fetch(no-cors) + median ms reported, four-second timeout.
Azure · 0 regions
// endpoint used
<region>.api.cognitive.microsoft.com — Cognitive Services regional subdomains. Unauth requests return fast, while TLS termination happens at the region's edge.
// what it measures
Each region gets 4 back-to-back samples; the first is treated as handshake warm-up and dropped, then the median of the remaining 3 is shown. Expect TCP + TLS + first-byte time — a touch above ICMP ping, with trustworthy ordering between regions.
// what it doesn't
No runs are stored, and your IP goes nowhere except to Microsoft's endpoints themselves. It also can't see region-to-region paths — for that, mtr from a VM inside Azure is the right tool.
// azure latency faq
How does this Azure latency test work?
The test times fetch(no-cors) requests from your browser to <region>.api.cognitive.microsoft.com — Cognitive Services subdomains that resolve and terminate TLS regionally, so the handshake genuinely travels to that region's network edge. Four sequential samples are taken per region; the first is discarded as connection warm-up and the median of the last three is shown. There is no backend and nothing is stored. What you see is roughly TCP plus TLS plus first byte, slightly above an ICMP ping.
How are Azure regions named?
Azure uses plain compass names with no AWS-style numbering scheme: eastus, westus2, northeurope, southeastasia, japaneast. Two traps catch newcomers. First, northeurope is actually Ireland and westeurope is the Netherlands — neither name tells you the country. Second, the digit in westus2 just distinguishes a separate location in the same geography. Display names like West Europe and resource-manager names like westeurope differ only by spacing and case, but the CLI and ARM templates want the lowercase form.
What are Azure region pairs?
Most Azure regions are statically paired with a second region in the same geopolitical area, usually hundreds of miles away: East US pairs with West US, North Europe with West Europe, Japan East with Japan West. Geo-redundant storage (GRS) replicates to the pair automatically, platform updates roll out to one side of a pair at a time, and after a major outage the pair gets recovery priority. Newer Azure regions increasingly skip pairing and lean on availability zones instead.
Do availability zones change latency inside an Azure region?
An Azure availability zone is a physically separate datacenter group within a region with independent power, cooling, and networking. Microsoft engineers zones to keep round-trip latency under roughly 2 ms, so zone-redundant deployments — zonal VMs behind a standard load balancer, ZRS storage, zone-redundant SQL — cost you almost nothing in latency while surviving the loss of a whole datacenter. This test measures to the region's network edge, so zone placement won't visibly move the numbers here.
What is Azure Front Door and when does it help?
Front Door is Azure's global anycast entry service: users connect to the nearest Microsoft edge PoP, TLS terminates there, and traffic rides Microsoft's private backbone to your origin region. Because the TCP and TLS handshakes happen close to the user, perceived latency drops even when the origin sits far away. So if your chosen region scores poorly in this test but your app is fronted by Front Door, real-user experience is likely better than the raw number suggests.
What are proximity placement groups?
This page measures user-to-region latency, but Azure also lets you squeeze VM-to-VM latency: a proximity placement group asks the platform to allocate your VMs physically close together — same datacenter, ideally the same network spine — pushing east-west latency down toward microsecond scale. It is the standard move for latency-sensitive clusters such as SAP, HPC, and trading workloads. The trade-off is placement flexibility: tight grouping can fail allocation during capacity crunches and concentrates your failure risk.
Which Azure region should I choose?
Start from this test run near your users, then check the product catalog: Azure's service availability varies by region more than most clouds, and popular regions periodically restrict new deployments due to capacity. eastus and westeurope have the broadest coverage; smaller regions may lack newer VM families or zone support entirely. Factor in your disaster-recovery story too — picking a region whose pair also suits your data-residency and latency needs saves real pain when you later add geo-redundancy.