Where Cloudflare Stands on Network Speed Heading into Fall
Cloudflare has benchmarked its edge network performance against other top global providers since June 2021, and it publishes regular updates on that progress. The latest data, measured from September 4 to September 19, 2024, shows Cloudflare ranking first in 44% of the top 1,000 networks as ranked by estimated population. That figure is based on 95th percentile (p95) TCP connection times — a metric that captures how long it takes an end user to establish a connection with the target endpoint.
The ranking count requires some clarification: Cloudflare is fastest in 410 of the measured networks, which would be 41% of 1,000. But the pool is trimmed to 932 networks after excluding transit providers that aren't last-mile ISPs, bringing the final share to 44%. That's consistent with the results Cloudflare reported in March 2024.
Looking at performance by country paints a similar picture, though the map shows fewer countries shaded as Cloudflare's territory compared to the March report.
That shift isn't necessarily a sign of regression. In many countries, the gap between the first- and second-fastest provider is under 5%, and in some cases the difference is under a millisecond. India is a current example where Cloudflare holds the top spot.
Rank | Entity | 95th percentile TCP Connect (ms) | Difference from #1 |
|---|---|---|---|
1 | Cloudflare | 290 ms | - |
2 | 291 ms | +0.28% (+0.81 ms) | |
3 | CloudFront | 304 ms | +4.61% (+13 ms) |
4 | Fastly | 325 ms | +12% (+35 ms) |
5 | Akamai | 373 ms | +29% (+83 ms) |
In India, Cloudflare edges out the runner-up by less than 1% — a margin so thin that the winner can change depending on which days the data is sampled. Network variance and last-mile outages can swing the numbers materially. For instance, a September 17, 2024 outage on a major Indian network caused a noticeable dip in Cloudflare's performance that day, as its automated systems aggressively rerouted traffic away from the affected provider due to detected packet loss. That response moved Cloudflare off the top spot for the day and has since prompted an internal review aimed at making outage-triggered traffic shifts more fine-grained.
Rank | Entity | 95th percentile TCP Connect (ms) | Difference from #1 |
|---|---|---|---|
1 | 219 ms | - | |
2 | CloudFront | 230 ms | +5% (+11 ms) |
3 | Cloudflare | 236 ms | +7.47% (+16 ms) |
4 | Fastly | 261 ms | +19% (+41 ms) |
5 | Akamai | 286 ms | +30% (+67 ms) |
Latency fluctuations can also work the other way. In Poland, Cloudflare was the fastest provider in March but has since slipped to second — though the difference is again less than a millisecond, suggesting the architecture remains competitive in the region.
Rank | Entity | 95th percentile TCP Connect (ms) | Difference from #1 |
|---|---|---|---|
1 | 246 ms | - | |
2 | Cloudflare | 246 ms | +0.15% (+0.36 ms) |
3 | CloudFront | 250 ms | +1.7% (+4.17 ms) |
4 | Akamai | 272 ms | +11% (+26 ms) |
5 | Fastly | 295 ms | +20% (+50 ms) |
These close margins can make Cloudflare appear slower in more countries than is truly the case. The company acknowledges the visualization of its performance data may need a longer look to more accurately represent areas where it is effectively tied with competitors.
How the Measurements Are Produced
Cloudflare's performance data comes from Real User Measurements (RUM) gathered via Cloudflare-branded error pages. When such a page is displayed, the browser fetches a small file from several major CDNs — Cloudflare, Akamai, Amazon CloudFront, Fastly, and Google Cloud CDN — and sends back metrics from the end user's perspective. This approach tracks TCP Connection Time, Time to First Byte (TTFB), and Time to Last Byte (TTLB) across those providers.
Since March 2024, Cloudflare has fixed the reference set of networks to the top 1,000 by estimated population (per APNIC data), excluding non-last-mile ISPs. That change standardizes the network pool across reporting cycles.
Using the Data to Shed Latency
This data isn't just for publishing rankings — it's used to identify and remediate underperforming regions. A 2022 example from Finland illustrates the workflow. At the time, users on one major Finnish network were experiencing p95 connection times above 300 ms. The root cause was congestion: Cloudflare was using all available capacity on that network path. After an expansion was completed, latency dropped within two weeks and Cloudflare became the fastest provider in the region.
Rank | Entity | 95th percentile TCP Connect (ms) | Difference from #1 |
|---|---|---|---|
1 | Fastly | 15 ms | - |
2 | CloudFront | 19 ms | +19% (+3 ms) |
3 | Akamai | 20 ms | +28% (+4.3 ms) |
4 | 72 ms | +363% (+56 ms) | |
5 | Cloudflare | 368 ms | +2378% (+353 ms) |
Cloudflare continuously tracks these metrics to locate where infrastructure investment will have the greatest impact on user-facing speed.
What Comes Next
The stated plan hasn't changed since the program began: find regions where Cloudflare lags, fix the underlying issue, and report on the improvement. With the tight margins now separating providers in many countries, the next challenge may be as much about measurement and visualization as it is about raw network tuning.
