Traditional speed tests often rely on ICMP packets. However, modern web traffic travels via HTTPS. GlobeLatency uses the Fetch API to execute actual requests against globally distributed assets. This provides a precise measurement of the Time to First Byte (TTFB) and total round-trip time.
The fundamental difference is this: ICMP ping tells you whether a server is reachable. GlobeLatency tells you how your connection performs when it matters — when loading a real website, calling a real API, or downloading a real file.
Our tool accounts for the time required to establish secure connections. This includes DNS resolution, TCP handshake, and the SSL/TLS negotiation, providing a realistic view of how long it takes for a user to actually "see" a website. The TLS handshake alone can add 50–150ms on a distant server — a cost that ICMP ping completely ignores.
To ensure 100% accuracy, we implement strict cache-busting protocols. Every request is unique, preventing ISPs, local proxies, or browser caches from returning "fake" high-speed results. We append a unique ?nocache=[timestamp] parameter to every URL, forcing a completely fresh server response on each measurement cycle. You see the raw, unbuffered performance of your line.
Many ISPs use a technique called "ICMP prioritization" — they detect ping packets and route them faster than regular traffic to inflate speed test scores. Some ISPs go further with "PowerBoost" — temporarily increasing bandwidth during the first few seconds of a test, then throttling back to normal. GlobeLatency uses HTTPS payloads that are indistinguishable from regular browsing traffic. There is no way for an ISP to selectively boost them without boosting all web traffic equally.
You might wonder: "Why not download a large file to test my speed?" Here is why our methodology focuses on small assets like favicons rather than large downloads.
Downloading a 1GB file measures throughput (bandwidth capacity), which tells you how much data can flow at once. However, latency is about reaction time. A 1GB transfer gets bogged down by congestion control and TCP slow-start, which masks the true physical latency of your connection.
By testing a 1KB favicon, the transfer completes almost instantly. This means the measured time is almost entirely composed of network RTT (Round Trip Time) and TTFB. It gives you a clean, "uncontaminated" view of your connection's speed, without being skewed by large data overhead.
Your browser doesn't wait for a 1GB file to load before showing you a website. It fetches small critical assets (CSS, JS, favicons) immediately. By testing these same small assets, GlobeLatency simulates the exact "handshake" your browser performs every day. It ensures our results reflect your actual experience when browsing the web, not just a theoretical maximum speed limit.
Your browser resolves the hostname to an IP address. This step reveals whether you have a fast DNS provider and whether the target uses GeoDNS to direct you to the nearest server.
A connection is established with the target server using a three-way handshake (SYN → SYN-ACK → ACK). This is the first physical round-trip to the server and reveals the base physical latency of your route.
For HTTPS targets, a cryptographic handshake is performed. Modern TLS 1.3 reduces this to a single round-trip, but older TLS 1.2 requires two. Our results include this real-world cost.
The exact millisecond the server starts sending its response is recorded. This is the most critical metric — it tells you how long a real user waits before they see anything on their screen.
Multiple cycles are performed per host. We calculate the average, minimum, and maximum values. A large difference between min and max reveals jitter — inconsistency that standard single-ping tests cannot detect.
"How is it possible that a Giant like Google or Steam responds in 20ms while my local shop's site takes 200ms?"
Huge platforms don't just use a CDN; they have direct "cables" (peering agreements) with your Internet Service Provider (ISP). When you request their favicon, the data travels from a server located potentially in the same building or street as your provider's main router. The same IP address resolves to different physical servers depending on your location — this is Anycast routing.
The 20ms you see is the time it takes to reach the Edge. The "Giant" has sliced its website into millions of tiny pieces and placed them right under your nose. You aren't loading the whole "Titan", you are just touching its "Skin" at the nearest point. Cloudflare alone operates over 300 edge locations globally — chances are, one is within 50km of you.
Imagine a global Pizza chain. Even if their headquarters is in Italy, they have a small kitchen in every neighborhood. When you order, the pizza doesn't fly from Italy; it comes from across the street. That is why it's hot and fast (20ms). GlobeLatency measures exactly how close those "kitchens" are to your current location — and whether your connection is routing to the local branch or accidentally sending your order to headquarters.
Use this table to interpret your GlobeLatency results:
| Latency Range | What It Means | Typical Cause | Impact |
|---|---|---|---|
| 0 – 10ms | Exceptional | Local CDN edge, direct ISP peering | Imperceptible — perfect for all use cases |
| 10 – 50ms | Excellent | National CDN node or nearby region | Ideal for gaming, video calls, trading |
| 50 – 120ms | Good | Continental routing, mid-tier CDN | Acceptable for most browsing and streaming |
| 120 – 250ms | Noticeable | Intercontinental routing, no CDN | Lag in games, delay in calls |
| 250ms+ | High / Problematic | Transoceanic cable, no CDN, routing issue | Significant lag — investigate ISP routing |
| TIMEOUT | Unreachable | Firewall, CORS block, server offline | Content not available from your location |
We don't test empty servers. We measure real production assets from the most critical infrastructure on the internet, grouped into categories that reveal different aspects of your connection quality.
Cloudflare, Akamai, and Fastly serve roughly 30% of all internet traffic. Testing against their edge nodes shows you how well your ISP is peered with the backbone of the modern web. AWS CloudFront, Google Cloud CDN, and Azure CDN reveal your latency to the world's largest cloud providers — critical for understanding SaaS application performance.
Steam (Valve), Epic Games, Riot Games (League of Legends), and Blizzard run their own private networks with direct ISP peering in most countries. Your latency to these endpoints directly predicts your in-game ping. If you see high latency to Steam here, your in-game ping will be similarly high regardless of your connection's download speed.
Cryptocurrency exchanges and financial platforms require ultra-low latency for time-sensitive operations. Testing against Binance, Coinbase, and similar platforms reveals whether your connection is competitive for trading, or whether ISP routing adds unnecessary milliseconds that could affect order execution.
University networks and national research institutes (like Tokyo University, MIT, and CERN) are connected directly to internet exchange points (IXPs). Testing these reveals your ISP's raw international routing quality — how efficiently your provider connects to the broader global internet without CDN assistance.
GlobeLatency deliberately tests small assets — favicons, tiny JavaScript files, and small CSS resources — rather than large files. Here is why: when you download a 100MB file, latency becomes irrelevant because the transfer time dominates the measurement. A 1KB favicon, however, transfers almost instantly once the connection is established. The measured time is almost entirely composed of network latency and server response time — exactly what we want to isolate. This gives you a clean, uncontaminated view of your connection's reaction speed.
We track the consistency of your connection. High jitter can be more damaging to video calls and gaming than high latency itself. A stable 80ms connection is superior to one that fluctuates between 20ms and 300ms. GlobeLatency's multi-cycle testing surfaces jitter that a single-ping test completely misses.
Instead of testing against "empty" servers, we ping real assets like Favicons, small CSS files, and JS snippets used in production environments. This simulates how your browser loads a real website — including all the overhead that standard network tests ignore.
Our engine handles Cross-Origin Resource Sharing (CORS) naturally, simulating exactly how a modern web application interacts with third-party APIs. Endpoints that block CORS are shown as timeouts — which is accurate, because they are effectively unreachable by browser-based applications.
By testing multiple servers in the same geographic region, GlobeLatency can reveal ISP routing inefficiencies. If two servers in the same country show dramatically different latencies, your ISP is taking different (and possibly suboptimal) routes to reach them.
Every test result is compared against our global average database, built from millions of measurements across all user locations. This tells you not just your latency, but whether your latency is better or worse than the global average for that specific server — giving your results real context.
Every audit generates a unique session report that can be shared with your ISP for technical support, or used as a baseline when comparing connections over time. The report includes per-host latency, global comparison, and your ISP fingerprint.
Through our Project 1 Dashboard, clients can monitor their own infrastructure. The dashboard provides a centralized place to:
This turns GlobeLatency from a simple test into a comprehensive network management suite — particularly useful for developers monitoring API response times, DevOps engineers tracking CDN performance, or businesses verifying that their hosting provider delivers consistent global performance.
Add your website's URL to the dashboard alongside URLs from competing hosting providers or CDNs. Run tests from your primary user location. The results will show you exactly how much latency improvement — or degradation — a hosting change would deliver to real users in your region. No marketing claims, no synthetic benchmarks — just measured real-world performance from your actual location.
Everything you need to know about network latency, how our tests work, and how to interpret your results.
Network latency is the time it takes for a data packet to travel from your device to a server and back — measured in milliseconds (ms). Low latency means a fast, responsive connection ideal for gaming, video calls, and trading. High latency causes lag, buffering, and slow page loads. Unlike download speed, latency measures reaction time — how quickly a server responds to your request, not how much data can flow at once. A connection with fast download speed but high latency will still feel slow and unresponsive.
Standard ping uses ICMP packets, which many ISPs artificially prioritize or fast-track to make connections appear faster than they are. GlobeLatency uses the browser's Fetch API to make real HTTPS requests — the same type of request your browser makes when loading a website. This means you see actual application-layer performance, not a synthetic network test that can be manipulated by traffic shaping. Our results reflect what real users actually experience.
Large platforms use Anycast IP routing and have direct peering agreements with major ISPs. This means their servers — or copies of their content — are physically located inside or very close to your ISP's network. When you request a Google asset, you are not connecting to a data center in California; you are connecting to a local edge node potentially just a few kilometers away. GlobeLatency measures exactly how close these edge nodes are to you — and shows when a site has poor CDN coverage in your region.
Under 30ms is excellent — you are hitting a local CDN edge node. 30–80ms is good for gaming and video calls. 80–150ms is average and acceptable for general browsing. 150–300ms indicates long physical distance or routing inefficiency. Above 300ms suggests international routing without CDN assistance, or packet loss on your line. Context also matters: 250ms to a Japanese server from Europe is normal and expected — light and data physically cannot travel faster than physics allows.
TTFB stands for Time to First Byte. It measures the moment from when your browser sends a request to when it receives the very first byte of the response. This is the most accurate measure of real-world server responsiveness — it captures DNS time, TCP handshake, TLS negotiation, server processing time, and the first network round-trip all in one number. GlobeLatency captures TTFB for every test, giving you the true reaction time of each server from your exact location.
Physical distance creates an unavoidable lower limit on latency. Light travels through fiber optic cables at roughly 200,000 km/s. The distance from Europe to Japan via undersea cables is approximately 20,000 km, meaning the absolute minimum round-trip time is around 100ms — and routing overhead adds more. 250ms to Japan from Europe means your connection is working correctly. If you see 10ms to Japan, you are hitting a local CDN cache, not Japan itself.
Cache busting means adding a unique timestamp to every request URL, like ?nocache=1715123456789. This forces the browser and any intermediate proxies to fetch a completely fresh copy of the file. Without cache busting, your browser might return a 1ms result from local memory — which tells you nothing about your actual network. GlobeLatency uses cache busting on every single request to guarantee real-time, unmanipulated measurements.
Yes. The GlobeLatency Dashboard allows you to add your own domains, APIs, or server endpoints for continuous monitoring. You can track your website's response time from your geographic location, compare it against global CDN benchmarks, and build custom monitoring templates. This is useful for developers and system administrators who want to monitor production infrastructure from a real-user perspective rather than from the server itself.
Mobile networks (4G and 5G) introduce additional processing overhead. Your device encodes data into radio waves, transmits to a cell tower, which then forwards through the carrier's core network. Even if your mobile download speed is high, the radio encoding and tower handshake add 30–80ms of base latency regardless. This is why fiber is preferred for latency-sensitive applications like online gaming, live trading, or real-time video collaboration — the physical medium itself is fundamentally faster.
Jitter is the variation in latency over multiple measurements. If your first request takes 20ms, the second 95ms, and the third 30ms, your jitter is high. High jitter destroys voice call quality, causes rubber-banding in online games, and makes video streams stutter. Consistent latency is often more important than low latency for these applications. GlobeLatency runs multiple test cycles per host specifically to expose jitter — a metric that single-ping tests completely miss.
ISP peering describes how your internet provider exchanges traffic with other networks. Direct peering means a short, efficient path. Without it, your traffic travels through multiple third-party transit networks, adding latency at each hop. GlobeLatency results can reveal bad peering: if your latency to a specific region is consistently high while neighboring regions are fast, your ISP likely routes that traffic through a longer or more congested path. This is actionable information you can bring to your ISP's technical support team.