Performance
Measured by someone else, every minute
Live
Median latency, last hour
United Kingdom, one-minute granularity. Hover the chart for exact values.
Current ranking
Median, last 15 min · UK · sample data
- 117.9ms
- 2Akamai20.5ms
- 3Cloudflare21.8ms
- 4CDN7722.3ms
- 5Amazon CloudFront22.5ms
- 6bunny.net22.7ms
- 7BlazingCDN24.8ms
- 8Google Cloud CDN27.0ms
Last 24 hours
—
Loading…
Last 7 days
—
Loading…
Last 30 days
—
Loading…
Independent real-user monitoring by PerfOps. Chart clipped at 32ms.
Independent real-user monitoring by PerfOps.
Why it's fast
Short trips, hot memory, a lean core
- Visitors reach the nearest of 988 locations, so most requests never cross an ocean
- Popular objects stay in RAM and the long tail sits on local NVMe
- The serving path is written in Rust, with async I/O and zero-copy sends
- RAM0.3ms
- NVMe1.8ms
- Origin84ms
Hot objects never leave memory
Same file, visitor in Manchester
Methodology
We don't mark our own homework
- Real User Monitoring from actual browsers, not synthetic probes in data centres
- A median per minute, so one bad connection can't skew the line
- Filterable by country and region; this page shows the UK
- The same eight providers and the same test object, every minute
01Real visitors
- ChromeManchester
- SafariLondon
- FirefoxLeeds
- EdgeBristol
Browsing ordinary sites that carry the PerfOps beacon.
02Same file, eight CDNs
Each browser times an identical object from every provider.
03Median per minute
- 18.4
- 19.1
- 61.2
- 18.9
- 22.7
19.1ms
One slow Wi-Fi hop can't drag the median.
04Published
- Ranked every minute
- Charted for the hour
- Summarised by day and month
Anyone can check the numbers on perfops.net.
Check it yourself
Time it from your own laptop
- x-cache shows HIT or MISS
- x-edge-location names the node that answered
- time_starttransfer is your time to first byte
curl -so /dev/null -D - https://cdn.example.com/app.js \
-w "ttfb: %{time_starttransfer}s\n" | grep -Ei "x-cache|x-edge|ttfb"
x-cache: HIT
x-edge-location: eu-west-1
ttfb: 0.011sFAQ
About the numbers
Who collects this data?
Why the United Kingdom?
Why median rather than average?
What if the live data doesn't load?
Does a few milliseconds matter?
Put your own traffic through it
Point one hostname at Edge, watch x-cache flip to HIT and compare the numbers with what you have today.
Free tier: 3 deployments and 500k requests a month, hard-capped. Bandwidth is always $0.